arize-phoenix 10.0.4__py3-none-any.whl → 12.28.1__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (276) hide show
  1. {arize_phoenix-10.0.4.dist-info → arize_phoenix-12.28.1.dist-info}/METADATA +124 -72
  2. arize_phoenix-12.28.1.dist-info/RECORD +499 -0
  3. {arize_phoenix-10.0.4.dist-info → arize_phoenix-12.28.1.dist-info}/WHEEL +1 -1
  4. {arize_phoenix-10.0.4.dist-info → arize_phoenix-12.28.1.dist-info}/licenses/IP_NOTICE +1 -1
  5. phoenix/__generated__/__init__.py +0 -0
  6. phoenix/__generated__/classification_evaluator_configs/__init__.py +20 -0
  7. phoenix/__generated__/classification_evaluator_configs/_document_relevance_classification_evaluator_config.py +17 -0
  8. phoenix/__generated__/classification_evaluator_configs/_hallucination_classification_evaluator_config.py +17 -0
  9. phoenix/__generated__/classification_evaluator_configs/_models.py +18 -0
  10. phoenix/__generated__/classification_evaluator_configs/_tool_selection_classification_evaluator_config.py +17 -0
  11. phoenix/__init__.py +5 -4
  12. phoenix/auth.py +39 -2
  13. phoenix/config.py +1763 -91
  14. phoenix/datetime_utils.py +120 -2
  15. phoenix/db/README.md +595 -25
  16. phoenix/db/bulk_inserter.py +145 -103
  17. phoenix/db/engines.py +140 -33
  18. phoenix/db/enums.py +3 -12
  19. phoenix/db/facilitator.py +302 -35
  20. phoenix/db/helpers.py +1000 -65
  21. phoenix/db/iam_auth.py +64 -0
  22. phoenix/db/insertion/dataset.py +135 -2
  23. phoenix/db/insertion/document_annotation.py +9 -6
  24. phoenix/db/insertion/evaluation.py +2 -3
  25. phoenix/db/insertion/helpers.py +17 -2
  26. phoenix/db/insertion/session_annotation.py +176 -0
  27. phoenix/db/insertion/span.py +15 -11
  28. phoenix/db/insertion/span_annotation.py +3 -4
  29. phoenix/db/insertion/trace_annotation.py +3 -4
  30. phoenix/db/insertion/types.py +50 -20
  31. phoenix/db/migrations/versions/01a8342c9cdf_add_user_id_on_datasets.py +40 -0
  32. phoenix/db/migrations/versions/0df286449799_add_session_annotations_table.py +105 -0
  33. phoenix/db/migrations/versions/272b66ff50f8_drop_single_indices.py +119 -0
  34. phoenix/db/migrations/versions/58228d933c91_dataset_labels.py +67 -0
  35. phoenix/db/migrations/versions/699f655af132_experiment_tags.py +57 -0
  36. phoenix/db/migrations/versions/735d3d93c33e_add_composite_indices.py +41 -0
  37. phoenix/db/migrations/versions/a20694b15f82_cost.py +196 -0
  38. phoenix/db/migrations/versions/ab513d89518b_add_user_id_on_dataset_versions.py +40 -0
  39. phoenix/db/migrations/versions/d0690a79ea51_users_on_experiments.py +40 -0
  40. phoenix/db/migrations/versions/deb2c81c0bb2_dataset_splits.py +139 -0
  41. phoenix/db/migrations/versions/e76cbd66ffc3_add_experiments_dataset_examples.py +87 -0
  42. phoenix/db/models.py +669 -56
  43. phoenix/db/pg_config.py +10 -0
  44. phoenix/db/types/model_provider.py +4 -0
  45. phoenix/db/types/token_price_customization.py +29 -0
  46. phoenix/db/types/trace_retention.py +23 -15
  47. phoenix/experiments/evaluators/utils.py +3 -3
  48. phoenix/experiments/functions.py +160 -52
  49. phoenix/experiments/tracing.py +2 -2
  50. phoenix/experiments/types.py +1 -1
  51. phoenix/inferences/inferences.py +1 -2
  52. phoenix/server/api/auth.py +38 -7
  53. phoenix/server/api/auth_messages.py +46 -0
  54. phoenix/server/api/context.py +100 -4
  55. phoenix/server/api/dataloaders/__init__.py +79 -5
  56. phoenix/server/api/dataloaders/annotation_configs_by_project.py +31 -0
  57. phoenix/server/api/dataloaders/annotation_summaries.py +60 -8
  58. phoenix/server/api/dataloaders/average_experiment_repeated_run_group_latency.py +50 -0
  59. phoenix/server/api/dataloaders/average_experiment_run_latency.py +17 -24
  60. phoenix/server/api/dataloaders/cache/two_tier_cache.py +1 -2
  61. phoenix/server/api/dataloaders/dataset_dataset_splits.py +52 -0
  62. phoenix/server/api/dataloaders/dataset_example_revisions.py +0 -1
  63. phoenix/server/api/dataloaders/dataset_example_splits.py +40 -0
  64. phoenix/server/api/dataloaders/dataset_examples_and_versions_by_experiment_run.py +47 -0
  65. phoenix/server/api/dataloaders/dataset_labels.py +36 -0
  66. phoenix/server/api/dataloaders/document_evaluation_summaries.py +2 -2
  67. phoenix/server/api/dataloaders/document_evaluations.py +6 -9
  68. phoenix/server/api/dataloaders/experiment_annotation_summaries.py +88 -34
  69. phoenix/server/api/dataloaders/experiment_dataset_splits.py +43 -0
  70. phoenix/server/api/dataloaders/experiment_error_rates.py +21 -28
  71. phoenix/server/api/dataloaders/experiment_repeated_run_group_annotation_summaries.py +77 -0
  72. phoenix/server/api/dataloaders/experiment_repeated_run_groups.py +57 -0
  73. phoenix/server/api/dataloaders/experiment_runs_by_experiment_and_example.py +44 -0
  74. phoenix/server/api/dataloaders/last_used_times_by_generative_model_id.py +35 -0
  75. phoenix/server/api/dataloaders/latency_ms_quantile.py +40 -8
  76. phoenix/server/api/dataloaders/record_counts.py +37 -10
  77. phoenix/server/api/dataloaders/session_annotations_by_session.py +29 -0
  78. phoenix/server/api/dataloaders/span_cost_by_span.py +24 -0
  79. phoenix/server/api/dataloaders/span_cost_detail_summary_entries_by_generative_model.py +56 -0
  80. phoenix/server/api/dataloaders/span_cost_detail_summary_entries_by_project_session.py +57 -0
  81. phoenix/server/api/dataloaders/span_cost_detail_summary_entries_by_span.py +43 -0
  82. phoenix/server/api/dataloaders/span_cost_detail_summary_entries_by_trace.py +56 -0
  83. phoenix/server/api/dataloaders/span_cost_details_by_span_cost.py +27 -0
  84. phoenix/server/api/dataloaders/span_cost_summary_by_experiment.py +57 -0
  85. phoenix/server/api/dataloaders/span_cost_summary_by_experiment_repeated_run_group.py +64 -0
  86. phoenix/server/api/dataloaders/span_cost_summary_by_experiment_run.py +58 -0
  87. phoenix/server/api/dataloaders/span_cost_summary_by_generative_model.py +55 -0
  88. phoenix/server/api/dataloaders/span_cost_summary_by_project.py +152 -0
  89. phoenix/server/api/dataloaders/span_cost_summary_by_project_session.py +56 -0
  90. phoenix/server/api/dataloaders/span_cost_summary_by_trace.py +55 -0
  91. phoenix/server/api/dataloaders/span_costs.py +29 -0
  92. phoenix/server/api/dataloaders/table_fields.py +2 -2
  93. phoenix/server/api/dataloaders/token_prices_by_model.py +30 -0
  94. phoenix/server/api/dataloaders/trace_annotations_by_trace.py +27 -0
  95. phoenix/server/api/dataloaders/types.py +29 -0
  96. phoenix/server/api/exceptions.py +11 -1
  97. phoenix/server/api/helpers/dataset_helpers.py +5 -1
  98. phoenix/server/api/helpers/playground_clients.py +1243 -292
  99. phoenix/server/api/helpers/playground_registry.py +2 -2
  100. phoenix/server/api/helpers/playground_spans.py +8 -4
  101. phoenix/server/api/helpers/playground_users.py +26 -0
  102. phoenix/server/api/helpers/prompts/conversions/aws.py +83 -0
  103. phoenix/server/api/helpers/prompts/conversions/google.py +103 -0
  104. phoenix/server/api/helpers/prompts/models.py +205 -22
  105. phoenix/server/api/input_types/{SpanAnnotationFilter.py → AnnotationFilter.py} +22 -14
  106. phoenix/server/api/input_types/ChatCompletionInput.py +6 -2
  107. phoenix/server/api/input_types/CreateProjectInput.py +27 -0
  108. phoenix/server/api/input_types/CreateProjectSessionAnnotationInput.py +37 -0
  109. phoenix/server/api/input_types/DatasetFilter.py +17 -0
  110. phoenix/server/api/input_types/ExperimentRunSort.py +237 -0
  111. phoenix/server/api/input_types/GenerativeCredentialInput.py +9 -0
  112. phoenix/server/api/input_types/GenerativeModelInput.py +5 -0
  113. phoenix/server/api/input_types/ProjectSessionSort.py +161 -1
  114. phoenix/server/api/input_types/PromptFilter.py +14 -0
  115. phoenix/server/api/input_types/PromptVersionInput.py +52 -1
  116. phoenix/server/api/input_types/SpanSort.py +44 -7
  117. phoenix/server/api/input_types/TimeBinConfig.py +23 -0
  118. phoenix/server/api/input_types/UpdateAnnotationInput.py +34 -0
  119. phoenix/server/api/input_types/UserRoleInput.py +1 -0
  120. phoenix/server/api/mutations/__init__.py +10 -0
  121. phoenix/server/api/mutations/annotation_config_mutations.py +8 -8
  122. phoenix/server/api/mutations/api_key_mutations.py +19 -23
  123. phoenix/server/api/mutations/chat_mutations.py +154 -47
  124. phoenix/server/api/mutations/dataset_label_mutations.py +243 -0
  125. phoenix/server/api/mutations/dataset_mutations.py +21 -16
  126. phoenix/server/api/mutations/dataset_split_mutations.py +351 -0
  127. phoenix/server/api/mutations/experiment_mutations.py +2 -2
  128. phoenix/server/api/mutations/export_events_mutations.py +3 -3
  129. phoenix/server/api/mutations/model_mutations.py +210 -0
  130. phoenix/server/api/mutations/project_mutations.py +49 -10
  131. phoenix/server/api/mutations/project_session_annotations_mutations.py +158 -0
  132. phoenix/server/api/mutations/project_trace_retention_policy_mutations.py +8 -4
  133. phoenix/server/api/mutations/prompt_label_mutations.py +74 -65
  134. phoenix/server/api/mutations/prompt_mutations.py +65 -129
  135. phoenix/server/api/mutations/prompt_version_tag_mutations.py +11 -8
  136. phoenix/server/api/mutations/span_annotations_mutations.py +15 -10
  137. phoenix/server/api/mutations/trace_annotations_mutations.py +14 -10
  138. phoenix/server/api/mutations/trace_mutations.py +47 -3
  139. phoenix/server/api/mutations/user_mutations.py +66 -41
  140. phoenix/server/api/queries.py +768 -293
  141. phoenix/server/api/routers/__init__.py +2 -2
  142. phoenix/server/api/routers/auth.py +154 -88
  143. phoenix/server/api/routers/ldap.py +229 -0
  144. phoenix/server/api/routers/oauth2.py +369 -106
  145. phoenix/server/api/routers/v1/__init__.py +24 -4
  146. phoenix/server/api/routers/v1/annotation_configs.py +23 -31
  147. phoenix/server/api/routers/v1/annotations.py +481 -17
  148. phoenix/server/api/routers/v1/datasets.py +395 -81
  149. phoenix/server/api/routers/v1/documents.py +142 -0
  150. phoenix/server/api/routers/v1/evaluations.py +24 -31
  151. phoenix/server/api/routers/v1/experiment_evaluations.py +19 -8
  152. phoenix/server/api/routers/v1/experiment_runs.py +337 -59
  153. phoenix/server/api/routers/v1/experiments.py +479 -48
  154. phoenix/server/api/routers/v1/models.py +7 -0
  155. phoenix/server/api/routers/v1/projects.py +18 -49
  156. phoenix/server/api/routers/v1/prompts.py +54 -40
  157. phoenix/server/api/routers/v1/sessions.py +108 -0
  158. phoenix/server/api/routers/v1/spans.py +1091 -81
  159. phoenix/server/api/routers/v1/traces.py +132 -78
  160. phoenix/server/api/routers/v1/users.py +389 -0
  161. phoenix/server/api/routers/v1/utils.py +3 -7
  162. phoenix/server/api/subscriptions.py +305 -88
  163. phoenix/server/api/types/Annotation.py +90 -23
  164. phoenix/server/api/types/ApiKey.py +13 -17
  165. phoenix/server/api/types/AuthMethod.py +1 -0
  166. phoenix/server/api/types/ChatCompletionSubscriptionPayload.py +1 -0
  167. phoenix/server/api/types/CostBreakdown.py +12 -0
  168. phoenix/server/api/types/Dataset.py +226 -72
  169. phoenix/server/api/types/DatasetExample.py +88 -18
  170. phoenix/server/api/types/DatasetExperimentAnnotationSummary.py +10 -0
  171. phoenix/server/api/types/DatasetLabel.py +57 -0
  172. phoenix/server/api/types/DatasetSplit.py +98 -0
  173. phoenix/server/api/types/DatasetVersion.py +49 -4
  174. phoenix/server/api/types/DocumentAnnotation.py +212 -0
  175. phoenix/server/api/types/Experiment.py +264 -59
  176. phoenix/server/api/types/ExperimentComparison.py +5 -10
  177. phoenix/server/api/types/ExperimentRepeatedRunGroup.py +155 -0
  178. phoenix/server/api/types/ExperimentRepeatedRunGroupAnnotationSummary.py +9 -0
  179. phoenix/server/api/types/ExperimentRun.py +169 -65
  180. phoenix/server/api/types/ExperimentRunAnnotation.py +158 -39
  181. phoenix/server/api/types/GenerativeModel.py +245 -3
  182. phoenix/server/api/types/GenerativeProvider.py +70 -11
  183. phoenix/server/api/types/{Model.py → InferenceModel.py} +1 -1
  184. phoenix/server/api/types/ModelInterface.py +16 -0
  185. phoenix/server/api/types/PlaygroundModel.py +20 -0
  186. phoenix/server/api/types/Project.py +1278 -216
  187. phoenix/server/api/types/ProjectSession.py +188 -28
  188. phoenix/server/api/types/ProjectSessionAnnotation.py +187 -0
  189. phoenix/server/api/types/ProjectTraceRetentionPolicy.py +1 -1
  190. phoenix/server/api/types/Prompt.py +119 -39
  191. phoenix/server/api/types/PromptLabel.py +42 -25
  192. phoenix/server/api/types/PromptVersion.py +11 -8
  193. phoenix/server/api/types/PromptVersionTag.py +65 -25
  194. phoenix/server/api/types/ServerStatus.py +6 -0
  195. phoenix/server/api/types/Span.py +167 -123
  196. phoenix/server/api/types/SpanAnnotation.py +189 -42
  197. phoenix/server/api/types/SpanCostDetailSummaryEntry.py +10 -0
  198. phoenix/server/api/types/SpanCostSummary.py +10 -0
  199. phoenix/server/api/types/SystemApiKey.py +65 -1
  200. phoenix/server/api/types/TokenPrice.py +16 -0
  201. phoenix/server/api/types/TokenUsage.py +3 -3
  202. phoenix/server/api/types/Trace.py +223 -51
  203. phoenix/server/api/types/TraceAnnotation.py +149 -50
  204. phoenix/server/api/types/User.py +137 -32
  205. phoenix/server/api/types/UserApiKey.py +73 -26
  206. phoenix/server/api/types/node.py +10 -0
  207. phoenix/server/api/types/pagination.py +11 -2
  208. phoenix/server/app.py +290 -45
  209. phoenix/server/authorization.py +38 -3
  210. phoenix/server/bearer_auth.py +34 -24
  211. phoenix/server/cost_tracking/cost_details_calculator.py +196 -0
  212. phoenix/server/cost_tracking/cost_model_lookup.py +179 -0
  213. phoenix/server/cost_tracking/helpers.py +68 -0
  214. phoenix/server/cost_tracking/model_cost_manifest.json +3657 -830
  215. phoenix/server/cost_tracking/regex_specificity.py +397 -0
  216. phoenix/server/cost_tracking/token_cost_calculator.py +57 -0
  217. phoenix/server/daemons/__init__.py +0 -0
  218. phoenix/server/daemons/db_disk_usage_monitor.py +214 -0
  219. phoenix/server/daemons/generative_model_store.py +103 -0
  220. phoenix/server/daemons/span_cost_calculator.py +99 -0
  221. phoenix/server/dml_event.py +17 -0
  222. phoenix/server/dml_event_handler.py +5 -0
  223. phoenix/server/email/sender.py +56 -3
  224. phoenix/server/email/templates/db_disk_usage_notification.html +19 -0
  225. phoenix/server/email/types.py +11 -0
  226. phoenix/server/experiments/__init__.py +0 -0
  227. phoenix/server/experiments/utils.py +14 -0
  228. phoenix/server/grpc_server.py +11 -11
  229. phoenix/server/jwt_store.py +17 -15
  230. phoenix/server/ldap.py +1449 -0
  231. phoenix/server/main.py +26 -10
  232. phoenix/server/oauth2.py +330 -12
  233. phoenix/server/prometheus.py +66 -6
  234. phoenix/server/rate_limiters.py +4 -9
  235. phoenix/server/retention.py +33 -20
  236. phoenix/server/session_filters.py +49 -0
  237. phoenix/server/static/.vite/manifest.json +55 -51
  238. phoenix/server/static/assets/components-BreFUQQa.js +6702 -0
  239. phoenix/server/static/assets/{index-E0M82BdE.js → index-CTQoemZv.js} +140 -56
  240. phoenix/server/static/assets/pages-DBE5iYM3.js +9524 -0
  241. phoenix/server/static/assets/vendor-BGzfc4EU.css +1 -0
  242. phoenix/server/static/assets/vendor-DCE4v-Ot.js +920 -0
  243. phoenix/server/static/assets/vendor-codemirror-D5f205eT.js +25 -0
  244. phoenix/server/static/assets/vendor-recharts-V9cwpXsm.js +37 -0
  245. phoenix/server/static/assets/vendor-shiki-Do--csgv.js +5 -0
  246. phoenix/server/static/assets/vendor-three-CmB8bl_y.js +3840 -0
  247. phoenix/server/templates/index.html +40 -6
  248. phoenix/server/thread_server.py +1 -2
  249. phoenix/server/types.py +14 -4
  250. phoenix/server/utils.py +74 -0
  251. phoenix/session/client.py +56 -3
  252. phoenix/session/data_extractor.py +5 -0
  253. phoenix/session/evaluation.py +14 -5
  254. phoenix/session/session.py +45 -9
  255. phoenix/settings.py +5 -0
  256. phoenix/trace/attributes.py +80 -13
  257. phoenix/trace/dsl/helpers.py +90 -1
  258. phoenix/trace/dsl/query.py +8 -6
  259. phoenix/trace/projects.py +5 -0
  260. phoenix/utilities/template_formatters.py +1 -1
  261. phoenix/version.py +1 -1
  262. arize_phoenix-10.0.4.dist-info/RECORD +0 -405
  263. phoenix/server/api/types/Evaluation.py +0 -39
  264. phoenix/server/cost_tracking/cost_lookup.py +0 -255
  265. phoenix/server/static/assets/components-DULKeDfL.js +0 -4365
  266. phoenix/server/static/assets/pages-Cl0A-0U2.js +0 -7430
  267. phoenix/server/static/assets/vendor-WIZid84E.css +0 -1
  268. phoenix/server/static/assets/vendor-arizeai-Dy-0mSNw.js +0 -649
  269. phoenix/server/static/assets/vendor-codemirror-DBtifKNr.js +0 -33
  270. phoenix/server/static/assets/vendor-oB4u9zuV.js +0 -905
  271. phoenix/server/static/assets/vendor-recharts-D-T4KPz2.js +0 -59
  272. phoenix/server/static/assets/vendor-shiki-BMn4O_9F.js +0 -5
  273. phoenix/server/static/assets/vendor-three-C5WAXd5r.js +0 -2998
  274. phoenix/utilities/deprecation.py +0 -31
  275. {arize_phoenix-10.0.4.dist-info → arize_phoenix-12.28.1.dist-info}/entry_points.txt +0 -0
  276. {arize_phoenix-10.0.4.dist-info → arize_phoenix-12.28.1.dist-info}/licenses/LICENSE +0 -0
phoenix/config.py CHANGED
@@ -1,5 +1,6 @@
1
1
  from __future__ import annotations
2
2
 
3
+ import json
3
4
  import logging
4
5
  import os
5
6
  import re
@@ -9,12 +10,25 @@ from datetime import timedelta
9
10
  from enum import Enum
10
11
  from importlib.metadata import version
11
12
  from pathlib import Path
12
- from typing import TYPE_CHECKING, Any, NamedTuple, Optional, Union, cast, overload
13
- from urllib.parse import quote_plus, urljoin, urlparse
13
+ from typing import (
14
+ TYPE_CHECKING,
15
+ Any,
16
+ Literal,
17
+ NamedTuple,
18
+ Optional,
19
+ TypedDict,
20
+ Union,
21
+ cast,
22
+ overload,
23
+ )
24
+ from urllib.parse import quote, urljoin, urlparse
14
25
 
15
26
  import wrapt
16
27
  from email_validator import EmailNotValidError, validate_email
28
+ from ldap3.core.exceptions import LDAPInvalidDnError
29
+ from ldap3.utils.dn import parse_dn
17
30
  from starlette.datastructures import URL, Secret
31
+ from typing_extensions import TypeAlias, get_args
18
32
 
19
33
  from phoenix.utilities.logging import log_a_list
20
34
  from phoenix.utilities.re import parse_env_headers
@@ -22,8 +36,17 @@ from phoenix.utilities.re import parse_env_headers
22
36
  if TYPE_CHECKING:
23
37
  from phoenix.server.oauth2 import OAuth2Clients
24
38
 
39
+ # Assignable roles (SYSTEM is internal-only and not included)
40
+ AssignableUserRoleName: TypeAlias = Literal["ADMIN", "MEMBER", "VIEWER"]
41
+
42
+ # Tuple of valid OAuth2 roles for validation
43
+ _VALID_ROLES: tuple[str, ...] = get_args(AssignableUserRoleName)
44
+
45
+
25
46
  logger = logging.getLogger(__name__)
26
47
 
48
+ ENV_OTEL_EXPORTER_OTLP_ENDPOINT = "OTEL_EXPORTER_OTLP_ENDPOINT"
49
+
27
50
  # Phoenix environment variables
28
51
  ENV_PHOENIX_PORT = "PHOENIX_PORT"
29
52
  ENV_PHOENIX_GRPC_PORT = "PHOENIX_GRPC_PORT"
@@ -50,6 +73,27 @@ ENV_PHOENIX_PROJECT_NAME = "PHOENIX_PROJECT_NAME"
50
73
  """
51
74
  The project name to use when logging traces and evals. defaults to 'default'.
52
75
  """
76
+ ENV_PHOENIX_FULLSTORY_ORG = "PHOENIX_FULLSTORY_ORG"
77
+ """
78
+ The FullStory organization ID for web analytics tracking. When set, FullStory tracking
79
+ will be enabled in the Phoenix web interface.
80
+ """
81
+ ENV_PHOENIX_SCARF_SH_PIXEL_ID = "PHOENIX_SCARF_SH_PIXEL_ID"
82
+ """
83
+ The Scarf.sh pixel ID for web analytics tracking. When set, Scarf.sh tracking
84
+ will be enabled in the Phoenix web interface.
85
+ """
86
+ ENV_PHOENIX_TELEMETRY_ENABLED = "PHOENIX_TELEMETRY_ENABLED"
87
+ """
88
+ Master toggle for telemetry pixels (FullStory and Scarf.sh).
89
+ When set to False, disables both FullStory and Scarf.sh tracking regardless of their
90
+ individual environment variable settings. Defaults to True.
91
+ """
92
+ ENV_PHOENIX_ALLOW_EXTERNAL_RESOURCES = "PHOENIX_ALLOW_EXTERNAL_RESOURCES"
93
+ """
94
+ Allows calls to external resources, like Google Fonts in the web interface
95
+ Defaults to True. Set to False in air-gapped environments to prevent external requests.
96
+ """
53
97
  ENV_PHOENIX_SQL_DATABASE_URL = "PHOENIX_SQL_DATABASE_URL"
54
98
  """
55
99
  The SQL database URL to use when logging traces and evals.
@@ -79,16 +123,44 @@ Used with PHOENIX_POSTGRES_HOST to specify the port to use for the PostgreSQL da
79
123
  ENV_PHOENIX_POSTGRES_USER = "PHOENIX_POSTGRES_USER"
80
124
  """
81
125
  Used with PHOENIX_POSTGRES_HOST to specify the user to use for the PostgreSQL database (required).
126
+
127
+ When using AWS RDS IAM authentication (PHOENIX_POSTGRES_USE_AWS_IAM_AUTH=true), this should be
128
+ set to the IAM-enabled database username configured in your RDS/Aurora instance.
82
129
  """
83
130
  ENV_PHOENIX_POSTGRES_PASSWORD = "PHOENIX_POSTGRES_PASSWORD"
84
131
  """
85
132
  Used with PHOENIX_POSTGRES_HOST to specify the password to use for the PostgreSQL database
86
- (required).
133
+ (required, unless PHOENIX_POSTGRES_USE_AWS_IAM_AUTH is enabled).
134
+
135
+ When using AWS RDS IAM authentication (PHOENIX_POSTGRES_USE_AWS_IAM_AUTH=true), this password
136
+ is NOT used. Instead, authentication tokens are generated dynamically using AWS IAM credentials.
87
137
  """
88
138
  ENV_PHOENIX_POSTGRES_DB = "PHOENIX_POSTGRES_DB"
89
139
  """
90
140
  Used with PHOENIX_POSTGRES_HOST to specify the database to use for the PostgreSQL database.
91
141
  """
142
+ ENV_PHOENIX_POSTGRES_USE_AWS_IAM_AUTH = "PHOENIX_POSTGRES_USE_AWS_IAM_AUTH"
143
+ """
144
+ Enable AWS RDS IAM database authentication. When enabled, Phoenix will use AWS IAM credentials
145
+ to generate short-lived authentication tokens instead of using a static password.
146
+
147
+ This requires:
148
+ - boto3 to be installed: pip install 'arize-phoenix[aws]'
149
+ - AWS credentials configured (via environment, ~/.aws/credentials, or IAM role)
150
+ - AWS region configured via standard AWS methods
151
+ - The database user to be configured for IAM authentication in RDS/Aurora
152
+ - SSL to be enabled (required by AWS RDS IAM auth)
153
+
154
+ When enabled, PHOENIX_POSTGRES_PASSWORD should NOT be set.
155
+ """
156
+ ENV_PHOENIX_POSTGRES_AWS_IAM_TOKEN_LIFETIME_SECONDS = (
157
+ "PHOENIX_POSTGRES_AWS_IAM_TOKEN_LIFETIME_SECONDS"
158
+ )
159
+ """
160
+ Token lifetime in seconds for connection pool recycling when using AWS RDS IAM authentication.
161
+ AWS RDS auth tokens are valid for 15 minutes. This should be set slightly lower to ensure
162
+ tokens are refreshed before expiration. Defaults to 840 seconds (14 minutes).
163
+ """
92
164
  ENV_PHOENIX_SQL_DATABASE_SCHEMA = "PHOENIX_SQL_DATABASE_SCHEMA"
93
165
  """
94
166
  The schema to use for the PostgresSQL database. (This is ignored for SQLite.)
@@ -99,12 +171,48 @@ ENV_PHOENIX_DATABASE_ALLOCATED_STORAGE_CAPACITY_GIBIBYTES = (
99
171
  )
100
172
  """
101
173
  The allocated storage capacity for the Phoenix database in gibibytes (2^30 bytes). Use float for
102
- fractional value. This is currently used only by the UI for informational displays.
174
+ fractional value.
175
+ """
176
+ ENV_PHOENIX_DATABASE_USAGE_EMAIL_WARNING_THRESHOLD_PERCENTAGE = (
177
+ "PHOENIX_DATABASE_USAGE_EMAIL_WARNING_THRESHOLD_PERCENTAGE"
178
+ )
179
+ """
180
+ The percentage of the allocated storage capacity that, when exceeded, triggers a email notifications
181
+ to admin users with valid email addresses. Must be specified in conjunction with allocated storage
182
+ capacity. This is a percentage value between 0 and 100. This setting is ignored if SMTP is not
183
+ configured.
184
+ """
185
+ ENV_PHOENIX_DATABASE_USAGE_INSERTION_BLOCKING_THRESHOLD_PERCENTAGE = (
186
+ "PHOENIX_DATABASE_USAGE_INSERTION_BLOCKING_THRESHOLD_PERCENTAGE"
187
+ )
188
+ """
189
+ The percentage of the allocated storage capacity that blocks insertions and updates of database
190
+ records when exceeded. Deletions are not blocked. Must be specified in conjunction with allocated
191
+ storage capacity. This is a percentage value between 0 and 100.
103
192
  """
104
193
  ENV_PHOENIX_ENABLE_PROMETHEUS = "PHOENIX_ENABLE_PROMETHEUS"
105
194
  """
106
195
  Whether to enable Prometheus. Defaults to false.
107
196
  """
197
+ ENV_PHOENIX_MAX_SPANS_QUEUE_SIZE = "PHOENIX_MAX_SPANS_QUEUE_SIZE"
198
+ """
199
+ The maximum number of spans to hold in the processing queue before rejecting new requests.
200
+
201
+ This is a heuristic to prevent memory issues when spans accumulate faster than they can be
202
+ written to the database. When this limit is reached, new incoming requests will be rejected
203
+ to protect system memory.
204
+
205
+ Note: The actual queue size may exceed this limit due to batch processing. Requests are
206
+ accepted or rejected before spans are deserialized, but a single accepted request may
207
+ contain multiple spans. This behavior is intentional to balance memory protection with
208
+ processing efficiency.
209
+
210
+ Memory usage: If an average span takes ~50KiB of memory, then 20,000 spans would use ~1GiB
211
+ of memory. Adjust this value based on your system's available memory and expected database
212
+ throughput.
213
+
214
+ Defaults to 20000.
215
+ """
108
216
  ENV_LOGGING_MODE = "PHOENIX_LOGGING_MODE"
109
217
  """
110
218
  The logging mode (either 'default' or 'structured').
@@ -128,8 +236,11 @@ ENV_PHOENIX_DANGEROUSLY_DISABLE_MIGRATIONS = "PHOENIX_DANGEROUSLY_DISABLE_MIGRAT
128
236
  """
129
237
  Whether or not to disable migrations. Defaults to None / False.
130
238
 
131
- This should only be used by developers working on the Phoenix server that need to be
132
- switching between branches without having to run migrations.
239
+ This should only be used by developers working on the Phoenix server that need
240
+ to be switching between branches without having to run migrations.
241
+
242
+ This can also be useful if a migration fails and you want to put the applicaiton
243
+ in a running state.
133
244
  """
134
245
 
135
246
  # Phoenix server OpenTelemetry instrumentation environment variables
@@ -140,6 +251,11 @@ ENV_PHOENIX_SERVER_INSTRUMENTATION_OTLP_TRACE_COLLECTOR_GRPC_ENDPOINT = (
140
251
  "PHOENIX_SERVER_INSTRUMENTATION_OTLP_TRACE_COLLECTOR_GRPC_ENDPOINT"
141
252
  )
142
253
 
254
+ ENV_PHOENIX_MASK_INTERNAL_SERVER_ERRORS = "PHOENIX_MASK_INTERNAL_SERVER_ERRORS"
255
+ """
256
+ Whether to mask internal server errors from the GraphQL and REST APIs. Defaults to true.
257
+ """
258
+
143
259
  # Authentication settings
144
260
  ENV_PHOENIX_ENABLE_AUTH = "PHOENIX_ENABLE_AUTH"
145
261
  ENV_PHOENIX_DISABLE_BASIC_AUTH = "PHOENIX_DISABLE_BASIC_AUTH"
@@ -190,15 +306,264 @@ password reset emails. If this variable is left unspecified or contains no origi
190
306
  protection will not be enabled. In such cases, when a request includes `origin` or `referer`
191
307
  headers, those values will not be validated.
192
308
  """
309
+
310
+ # LDAP authentication settings
311
+ ENV_PHOENIX_LDAP_HOST = "PHOENIX_LDAP_HOST"
312
+ """
313
+ LDAP server hosts (comma-separated for multiple servers with failover).
314
+ Example: "ldap.corp.com" or "dc1.corp.com,dc2.corp.com,dc3.corp.com"
315
+
316
+ Multi-server failover behavior:
317
+ - Connection errors (server unreachable, timeout): Automatically tries the next server
318
+ - User not found: Returns immediately (no failover to other servers)
319
+ - Invalid password: Returns immediately (no failover to other servers)
320
+
321
+ This assumes all servers are replicas with identical user sets (the common HA pattern).
322
+ Multi-domain/forest configurations where different users exist on different servers
323
+ are NOT supported.
324
+ """
325
+ ENV_PHOENIX_LDAP_PORT = "PHOENIX_LDAP_PORT"
326
+ """
327
+ LDAP server port. Defaults to 389 for StartTLS, 636 for LDAPS.
328
+ """
329
+ ENV_PHOENIX_LDAP_TLS_MODE = "PHOENIX_LDAP_TLS_MODE"
330
+ """
331
+ TLS connection mode. Defaults to "starttls". Options:
332
+ - "starttls": Upgrade from plaintext to TLS on port 389 (recommended)
333
+ - "ldaps": TLS from connection start on port 636
334
+ - "none": No encryption (testing only, credentials sent in plaintext)
335
+ """
336
+ ENV_PHOENIX_LDAP_TLS_VERIFY = "PHOENIX_LDAP_TLS_VERIFY"
337
+ """
338
+ Verify TLS certificates. Defaults to true. Should always be true in production.
339
+ """
340
+ ENV_PHOENIX_LDAP_TLS_CA_CERT_FILE = "PHOENIX_LDAP_TLS_CA_CERT_FILE"
341
+ """
342
+ Path to custom CA certificate file (PEM format) for TLS verification. Optional.
343
+ Use when LDAP server uses a private/internal CA not in the system trust store.
344
+ Example: "/etc/ssl/certs/internal-ca.pem"
345
+ """
346
+ ENV_PHOENIX_LDAP_TLS_CLIENT_CERT_FILE = "PHOENIX_LDAP_TLS_CLIENT_CERT_FILE"
347
+ """
348
+ Path to client certificate file (PEM format) for mutual TLS authentication. Optional.
349
+ Requires PHOENIX_LDAP_TLS_CLIENT_KEY_FILE to also be set.
350
+ Example: "/etc/ssl/certs/phoenix-client.crt"
351
+ """
352
+ ENV_PHOENIX_LDAP_TLS_CLIENT_KEY_FILE = "PHOENIX_LDAP_TLS_CLIENT_KEY_FILE"
353
+ """
354
+ Path to client private key file (PEM format) for mutual TLS authentication. Optional.
355
+ Requires PHOENIX_LDAP_TLS_CLIENT_CERT_FILE to also be set.
356
+ Example: "/etc/ssl/private/phoenix-client.key"
357
+ """
358
+ ENV_PHOENIX_LDAP_BIND_DN = "PHOENIX_LDAP_BIND_DN"
359
+ """
360
+ Service account DN for binding to LDAP server. Optional for direct bind.
361
+ Example: "CN=svc-phoenix,OU=Service Accounts,DC=corp,DC=com"
362
+ """
363
+ ENV_PHOENIX_LDAP_BIND_PASSWORD = "PHOENIX_LDAP_BIND_PASSWORD"
364
+ """
365
+ Service account password for binding to LDAP server.
366
+ Required if BIND_DN is set. Should be stored securely (e.g., Kubernetes Secret,
367
+ environment variable from secrets manager). Avoid hardcoding in configuration files.
368
+ """
369
+ ENV_PHOENIX_LDAP_USER_SEARCH_BASE_DNS = "PHOENIX_LDAP_USER_SEARCH_BASE_DNS"
370
+ """
371
+ JSON array of base DNs for user searches. Searches are performed in order until a user is found.
372
+ Example: '["OU=Users,DC=corp,DC=com"]'
373
+ Multiple: '["OU=Employees,DC=corp,DC=com", "OU=Contractors,DC=corp,DC=com"]'
374
+ """
375
+ ENV_PHOENIX_LDAP_USER_SEARCH_FILTER = "PHOENIX_LDAP_USER_SEARCH_FILTER"
376
+ """
377
+ LDAP filter for finding users. Use %s as placeholder for username.
378
+ Example: "(&(objectClass=user)(sAMAccountName=%s))"
379
+ """
380
+ ENV_PHOENIX_LDAP_ATTR_EMAIL = "PHOENIX_LDAP_ATTR_EMAIL"
381
+ """
382
+ LDAP attribute containing user's email address. Required.
383
+
384
+ Set to the attribute name (e.g., "mail") if your LDAP directory has email addresses.
385
+ The attribute must be present in LDAP or login fails.
386
+
387
+ Set to "null" if your LDAP directory does not have email addresses. When set to "null":
388
+ - PHOENIX_LDAP_ATTR_UNIQUE_ID is required (users are identified by unique_id instead)
389
+ - PHOENIX_LDAP_ALLOW_SIGN_UP must be true (users are auto-provisioned on first login)
390
+ - PHOENIX_ADMINS is not supported (use PHOENIX_LDAP_GROUP_ROLE_MAPPINGS instead)
391
+ - Users will appear in Phoenix without email addresses
392
+
393
+ https://www.rfc-editor.org/rfc/rfc2798#section-9.1.3
394
+ """
395
+ ENV_PHOENIX_LDAP_ATTR_UNIQUE_ID = "PHOENIX_LDAP_ATTR_UNIQUE_ID"
396
+ """
397
+ LDAP attribute containing an immutable unique identifier.
398
+
399
+ REQUIRED when PHOENIX_LDAP_ATTR_EMAIL is "null" (users are identified by this ID
400
+ instead of email).
401
+
402
+ Also recommended if you expect user emails to change (company rebranding, M&A,
403
+ frequent name changes) or have compliance requirements for immutable user tracking.
404
+ For most organizations with email in LDAP, the default email-based identification
405
+ is sufficient.
406
+
407
+ When set, this attribute is used as the primary identifier, allowing users
408
+ to survive email changes without creating duplicate accounts.
409
+
410
+ Supported attributes (UUID-based only):
411
+ - Active Directory: "objectGUID" (16-byte binary UUID)
412
+ - OpenLDAP: "entryUUID" (RFC 4530, string UUID)
413
+ - 389 Directory Server: "nsUniqueId" (string UUID)
414
+
415
+ IMPORTANT: Only standard UUID-based attributes are supported. Custom attributes
416
+ containing 16-character string IDs (e.g., "EMP12345ABCD6789") are NOT supported
417
+ and will be incorrectly converted.
418
+
419
+ When not set (default), email is used as the identifier. Both modes handle
420
+ DN changes (OU moves, renames). The only difference is email change handling.
421
+ """
422
+ ENV_PHOENIX_LDAP_ATTR_DISPLAY_NAME = "PHOENIX_LDAP_ATTR_DISPLAY_NAME"
423
+ """
424
+ LDAP attribute containing user's display name. Defaults to "displayName".
425
+ https://www.rfc-editor.org/rfc/rfc2798.html#section-2.3
426
+ """
427
+ ENV_PHOENIX_LDAP_ATTR_MEMBER_OF = "PHOENIX_LDAP_ATTR_MEMBER_OF"
428
+ """
429
+ LDAP attribute containing group memberships. Defaults to "memberOf".
430
+ Used for Active Directory and OpenLDAP with memberOf overlay.
431
+ This attribute is only used when GROUP_SEARCH_FILTER is not set.
432
+ """
433
+ ENV_PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS = "PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS"
434
+ """
435
+ JSON array of base DNs for group searches (for POSIX/OpenLDAP).
436
+ Required when using GROUP_SEARCH_FILTER.
437
+ Example: '["ou=groups,dc=example,dc=com"]'
438
+ Multiple: '["ou=groups,dc=corp,dc=com", "ou=teams,dc=corp,dc=com"]'
439
+ """
440
+ ENV_PHOENIX_LDAP_GROUP_SEARCH_FILTER = "PHOENIX_LDAP_GROUP_SEARCH_FILTER"
441
+ """
442
+ LDAP filter for finding groups containing a user. Use %s as placeholder for user identifier.
443
+
444
+ Two Group Resolution Modes
445
+ --------------------------
446
+ This setting determines how group membership is resolved:
447
+
448
+ 1. AD Mode (this setting NOT set - RECOMMENDED for Active Directory):
449
+ Reads the memberOf attribute directly from the user entry.
450
+ Active Directory automatically populates this attribute.
451
+ Configure PHOENIX_LDAP_ATTR_MEMBER_OF if the attribute name differs.
452
+
453
+ 2. Search Mode (this setting IS set):
454
+ Searches for groups that contain the user in PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS.
455
+ Required for POSIX groups (posixGroup) or when memberOf is unavailable.
456
+
457
+ Placeholder Substitution
458
+ ------------------------
459
+ The %s placeholder is replaced with a user identifier. What value is used depends on
460
+ PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR:
461
+ - If not set (default): Uses the login username directly
462
+ - If set: Uses that attribute's value from the user's LDAP entry
463
+
464
+ See PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR for detailed examples of:
465
+ - POSIX groups (memberUid contains usernames)
466
+ - groupOfNames (member contains full DNs)
467
+
468
+ Example POSIX configuration:
469
+ PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS=["ou=groups,dc=example,dc=com"]
470
+ PHOENIX_LDAP_GROUP_SEARCH_FILTER=(&(objectClass=posixGroup)(memberUid=%s))
471
+ # PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR not needed - uses login username
472
+ """
473
+ ENV_PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR = "PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR"
474
+ """
475
+ LDAP attribute from the user entry to substitute for %s in GROUP_SEARCH_FILTER.
476
+
477
+ When set, reads the specified attribute from the user's LDAP entry and uses its value.
478
+ When not set (default), uses the login username directly.
479
+
480
+ Understanding Group Membership Attributes
481
+ -----------------------------------------
482
+ Different LDAP group types store membership differently:
483
+
484
+ 1. POSIX groups (posixGroup objectClass):
485
+ - Use "memberUid" attribute which contains **usernames** (e.g., "jdoe")
486
+ - Filter: (&(objectClass=posixGroup)(memberUid=%s))
487
+ - Use: PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR not set (uses login username)
488
+ or set to "uid" if login username differs from uid attribute
489
+
490
+ 2. groupOfNames/groupOfUniqueNames (RFC 4519):
491
+ - Use "member"/"uniqueMember" which contains **full DNs**
492
+ (e.g., "uid=jdoe,ou=users,dc=example,dc=com")
493
+ - Filter: (&(objectClass=groupOfNames)(member=%s))
494
+ - Use: PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR=distinguishedName (AD only)
495
+ - Note: OpenLDAP does not expose DN as an attribute. For groupOfNames with
496
+ OpenLDAP, consider using memberOf overlay instead (AD mode).
497
+
498
+ 3. Active Directory groups:
499
+ - RECOMMENDED: Use AD mode (memberOf attribute) instead of group search.
500
+ AD automatically populates memberOf on user entries.
501
+ Simply leave PHOENIX_LDAP_GROUP_SEARCH_FILTER unset.
502
+ - If you must use group search: AD returns "distinguishedName" as an attribute,
503
+ so PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR=distinguishedName works.
504
+
505
+ Common values:
506
+ - Not set (default): Uses the login username directly
507
+ - "uid": Explicitly use uid attribute (same as default for most setups)
508
+ - "distinguishedName": Full DN (Active Directory only)
509
+ - "sAMAccountName": Windows login name (Active Directory)
510
+
511
+ Example configurations:
512
+
513
+ POSIX groups with OpenLDAP (memberUid contains usernames):
514
+ PHOENIX_LDAP_GROUP_SEARCH_FILTER=(&(objectClass=posixGroup)(memberUid=%s))
515
+ # No PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR needed - uses login username
516
+
517
+ POSIX groups when login differs from uid:
518
+ PHOENIX_LDAP_GROUP_SEARCH_FILTER=(&(objectClass=posixGroup)(memberUid=%s))
519
+ PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR=uid
520
+
521
+ Active Directory with group search (not recommended - use memberOf instead):
522
+ PHOENIX_LDAP_GROUP_SEARCH_FILTER=(member:1.2.840.113556.1.4.1941:=%s)
523
+ PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR=distinguishedName
524
+ """
525
+ ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS = "PHOENIX_LDAP_GROUP_ROLE_MAPPINGS"
526
+ """
527
+ JSON array mapping LDAP groups to Phoenix roles.
528
+ Example: '[{"group_dn": "CN=Phoenix Admins,OU=Groups,DC=corp,DC=com", "role": "ADMIN"}]'
529
+ Supported role values: "ADMIN", "MEMBER", "VIEWER" (case-insensitive).
530
+ Special group_dn value "*" matches all users (wildcard).
531
+ Order matters: first matching group_dn in the array determines the role.
532
+ """
533
+ ENV_PHOENIX_LDAP_ALLOW_SIGN_UP = "PHOENIX_LDAP_ALLOW_SIGN_UP"
534
+ """
535
+ Allow automatic user creation on first LDAP login. Defaults to "true".
536
+ Set to "false" to require pre-provisioned users (created via PHOENIX_ADMINS
537
+ env var or the application's user management UI before first login).
538
+ Pre-provisioned users are matched by email on first LDAP login.
539
+
540
+ MUST be "true" when PHOENIX_LDAP_ATTR_EMAIL is "null", since pre-provisioning
541
+ by email is not possible without email addresses in LDAP.
542
+ """
543
+
193
544
  ENV_PHOENIX_ADMINS = "PHOENIX_ADMINS"
194
545
  """
195
546
  A semicolon-separated list of username and email address pairs to create as admin users on startup.
196
547
  The format is `username=email`, e.g., `John Doe=john@example.com;Doe, Jane=jane@example.com`.
197
- The password for each user will be randomly generated and will need to be reset. The application
198
- will not start if this environment variable is set but cannot be parsed or contains invalid emails.
199
- If the username or email address already exists in the database, the user record will not be
200
- modified, e.g., changed from non-admin to admin. Changing this environment variable for the next
201
- startup will not undo any records created in previous startups.
548
+
549
+ User Type Selection:
550
+ The type of user created depends on the authentication configuration:
551
+ - If basic auth is enabled (PHOENIX_DISABLE_BASIC_AUTH=false, default): Creates a LocalUser
552
+ with a randomly generated password that must be reset on first login.
553
+ - If basic auth is disabled AND LDAP is configured: Creates an LDAPUser that will be matched
554
+ by email on first LDAP login.
555
+ - If basic auth is disabled AND only OAuth2 is configured: Creates an OAuth2User that will be
556
+ matched by email on first OAuth2 login.
557
+
558
+ Notes:
559
+ - The application will not start if this environment variable is set but cannot be parsed or
560
+ contains invalid emails.
561
+ - If the username or email address already exists in the database, the user record will not be
562
+ modified, e.g., changed from non-admin to admin.
563
+ - Changing this environment variable for the next startup will not undo any records created in
564
+ previous startups.
565
+ - NOT supported when PHOENIX_LDAP_ATTR_EMAIL is "null" (use PHOENIX_LDAP_GROUP_ROLE_MAPPINGS
566
+ to assign admin roles instead when LDAP doesn't have email addresses).
202
567
  """
203
568
  ENV_PHOENIX_ROOT_URL = "PHOENIX_ROOT_URL"
204
569
  """
@@ -213,7 +578,20 @@ Examples:
213
578
  - With a sub-path: "https://example.com/phoenix"
214
579
  - Without a sub-path: "https://phoenix.example.com"
215
580
  """
581
+ ENV_PHOENIX_MANAGEMENT_URL = "PHOENIX_MANAGEMENT_URL"
582
+ """
583
+ The URL to use for redirecting to a management interface that may be hosting Phoenix. If set, and
584
+ the current user is within PHOENIX_ADMINS, a link will be added to the navigation menu to return to
585
+ this URL.
586
+ """
587
+ ENV_PHOENIX_SUPPORT_EMAIL = "PHOENIX_SUPPORT_EMAIL"
588
+ """
589
+ The support email address to display in error messages and notifications.
216
590
 
591
+ When set, this email will be included in error messages for insufficient storage
592
+ conditions and database usage notification emails, providing users with a direct
593
+ contact for assistance. If not set, error messages will not include contact information.
594
+ """
217
595
 
218
596
  # SMTP settings
219
597
  ENV_PHOENIX_SMTP_HOSTNAME = "PHOENIX_SMTP_HOSTNAME"
@@ -288,6 +666,10 @@ Whether to verify client certificates for mutual TLS (mTLS) authentication.
288
666
  When set to true, clients must provide valid certificates signed by the CA specified in
289
667
  PHOENIX_TLS_CA_FILE.
290
668
  """
669
+ ENV_PHOENIX_DEFAULT_RETENTION_POLICY_DAYS = "PHOENIX_DEFAULT_RETENTION_POLICY_DAYS"
670
+ """
671
+ The default retention policy for traces in days.
672
+ """
291
673
 
292
674
 
293
675
  @dataclass(frozen=True)
@@ -449,6 +831,20 @@ def get_env_tls_verify_client() -> bool:
449
831
  return _bool_val(ENV_PHOENIX_TLS_VERIFY_CLIENT, False)
450
832
 
451
833
 
834
+ def get_env_default_retention_policy_days() -> int:
835
+ """
836
+ Returns the number of days for the default retention policy as set by the
837
+ PHOENIX_DEFAULT_RETENTION_POLICY_DAYS environment variable, defaulting to 0 if not set.
838
+
839
+ Returns:
840
+ int: Number of days for the default retention policy. Defaults to 0 if the environment variable is not set.
841
+ """ # noqa: E501
842
+ days = _int_val(ENV_PHOENIX_DEFAULT_RETENTION_POLICY_DAYS, 0)
843
+ if days < 0:
844
+ raise ValueError("PHOENIX_DEFAULT_RETENTION_POLICY_DAYS must be non-negative")
845
+ return days
846
+
847
+
452
848
  def get_env_tls_config() -> Optional[TLSConfig]:
453
849
  """
454
850
  Retrieves and validates TLS configuration from environment variables.
@@ -717,6 +1113,7 @@ class AuthSettings(NamedTuple):
717
1113
  phoenix_secret: Secret
718
1114
  phoenix_admin_secret: Secret
719
1115
  oauth2_clients: OAuth2Clients
1116
+ ldap_config: Optional[LDAPConfig]
720
1117
 
721
1118
 
722
1119
  def get_env_auth_settings() -> AuthSettings:
@@ -735,9 +1132,13 @@ def get_env_auth_settings() -> AuthSettings:
735
1132
  from phoenix.server.oauth2 import OAuth2Clients
736
1133
 
737
1134
  oauth2_clients = OAuth2Clients.from_configs(get_env_oauth2_settings())
738
- if enable_auth and disable_basic_auth and not oauth2_clients:
1135
+ ldap_config = LDAPConfig.from_env()
1136
+
1137
+ if enable_auth and disable_basic_auth and not oauth2_clients and not ldap_config:
739
1138
  raise ValueError(
740
- "OAuth2 is the only supported auth method but no OAuth2 client configs are provided."
1139
+ f"{ENV_PHOENIX_DISABLE_BASIC_AUTH} is set, but no alternative authentication methods "
1140
+ "are configured. Please configure at least one of: OAuth2 "
1141
+ f"(PHOENIX_OAUTH2_*) or LDAP ({ENV_PHOENIX_LDAP_HOST})."
741
1142
  )
742
1143
  return AuthSettings(
743
1144
  enable_auth=enable_auth,
@@ -745,6 +1146,7 @@ def get_env_auth_settings() -> AuthSettings:
745
1146
  phoenix_secret=phoenix_secret,
746
1147
  phoenix_admin_secret=phoenix_admin_secret,
747
1148
  oauth2_clients=oauth2_clients,
1149
+ ldap_config=ldap_config,
748
1150
  )
749
1151
 
750
1152
 
@@ -808,7 +1210,7 @@ def get_env_csrf_trusted_origins() -> list[str]:
808
1210
 
809
1211
  def get_env_admins() -> dict[str, str]:
810
1212
  """
811
- Parse the PHOENIX_ADMINS environment variable to extract the comma separated pairs of
1213
+ Parse the PHOENIX_ADMINS environment variable to extract the semicolon separated pairs of
812
1214
  username and email. The last equal sign (=) in each pair is used to separate the username from
813
1215
  the email.
814
1216
 
@@ -820,6 +1222,8 @@ def get_env_admins() -> dict[str, str]:
820
1222
  """
821
1223
  if not (env_value := getenv(ENV_PHOENIX_ADMINS)):
822
1224
  return {}
1225
+ from phoenix.auth import sanitize_email
1226
+
823
1227
  usernames = set()
824
1228
  emails = set()
825
1229
  ans = {}
@@ -836,7 +1240,7 @@ def get_env_admins() -> dict[str, str]:
836
1240
  f"Expected format: 'username=email'"
837
1241
  )
838
1242
  username = pair[:last_equals_pos].strip()
839
- email_addr = pair[last_equals_pos + 1 :].strip()
1243
+ email_addr = sanitize_email(pair[last_equals_pos + 1 :])
840
1244
  try:
841
1245
  email_addr = validate_email(email_addr, check_deliverability=False).normalized
842
1246
  except EmailNotValidError:
@@ -877,90 +1281,1056 @@ def get_env_smtp_validate_certs() -> bool:
877
1281
  return _bool_val(ENV_PHOENIX_SMTP_VALIDATE_CERTS, True)
878
1282
 
879
1283
 
1284
+ _ALLOWED_TOKEN_ENDPOINT_AUTH_METHODS = (
1285
+ "client_secret_basic",
1286
+ "client_secret_post",
1287
+ "none",
1288
+ )
1289
+ """Allowed OAuth2 token endpoint authentication methods (OIDC Core §9)."""
1290
+
1291
+
880
1292
  @dataclass(frozen=True)
881
1293
  class OAuth2ClientConfig:
1294
+ """Configuration for an OAuth2/OIDC identity provider."""
1295
+
1296
+ # Identity provider identification
882
1297
  idp_name: str
883
1298
  idp_display_name: str
1299
+
1300
+ # OAuth2 client credentials (RFC 6749 §2)
884
1301
  client_id: str
885
- client_secret: str
1302
+ client_secret: Optional[
1303
+ str
1304
+ ] # Optional when token_endpoint_auth_method is "none" (RFC 6749 §2.3.1)
886
1305
  oidc_config_url: str
1306
+
1307
+ # Authentication behavior
887
1308
  allow_sign_up: bool
888
1309
  auto_login: bool
1310
+ use_pkce: bool # Proof Key for Code Exchange (RFC 7636)
1311
+ token_endpoint_auth_method: Optional[str] # OIDC Core §9
1312
+
1313
+ # Scopes and permissions (RFC 6749 §3.3: space-delimited)
1314
+ scopes: str
1315
+
1316
+ # Group-based access control
1317
+ groups_attribute_path: Optional[str]
1318
+ allowed_groups: list[str]
1319
+
1320
+ # Role mapping
1321
+ role_attribute_path: Optional[str]
1322
+ role_mapping: dict[str, AssignableUserRoleName]
1323
+ role_attribute_strict: bool
889
1324
 
890
1325
  @classmethod
891
1326
  def from_env(cls, idp_name: str) -> "OAuth2ClientConfig":
892
- idp_name_upper = idp_name.upper()
893
- if not (
894
- client_id := getenv(client_id_env_var := f"PHOENIX_OAUTH2_{idp_name_upper}_CLIENT_ID")
895
- ):
896
- raise ValueError(
897
- f"A client id must be set for the {idp_name} OAuth2 IDP "
898
- f"via the {client_id_env_var} environment variable"
899
- )
900
- if not (
901
- client_secret := getenv(
902
- client_secret_env_var := f"PHOENIX_OAUTH2_{idp_name_upper}_CLIENT_SECRET"
903
- )
904
- ):
905
- raise ValueError(
906
- f"A client secret must be set for the {idp_name} OAuth2 IDP "
907
- f"via the {client_secret_env_var} environment variable"
908
- )
909
- if not (
910
- oidc_config_url := (
911
- getenv(
912
- oidc_config_url_env_var := f"PHOENIX_OAUTH2_{idp_name_upper}_OIDC_CONFIG_URL",
1327
+ """Load OAuth2 client configuration from environment variables for the given IDP name."""
1328
+ idp_prefix = f"PHOENIX_OAUTH2_{idp_name.upper()}"
1329
+
1330
+ def _get_required(suffix: str, description: str) -> str:
1331
+ """Get a required environment variable or raise a descriptive error."""
1332
+ env_var = f"{idp_prefix}_{suffix}"
1333
+ value = getenv(env_var)
1334
+ if value is None or not value:
1335
+ raise ValueError(
1336
+ f"{description} must be set for the {idp_name} OAuth2 IDP "
1337
+ f"via the {env_var} environment variable"
913
1338
  )
1339
+ return value
1340
+
1341
+ def _get_optional(suffix: str) -> Optional[str]:
1342
+ """Get an optional environment variable."""
1343
+ return getenv(f"{idp_prefix}_{suffix}")
1344
+
1345
+ # Required configuration
1346
+ client_id = _get_required("CLIENT_ID", "Client ID")
1347
+ oidc_config_url = _get_required("OIDC_CONFIG_URL", "OpenID Connect configuration URL")
1348
+
1349
+ # Validate OIDC URL format and HTTPS requirement
1350
+ parsed_url = urlparse(oidc_config_url)
1351
+ if not parsed_url.scheme or not parsed_url.hostname:
1352
+ raise ValueError(
1353
+ f"Invalid OIDC configuration URL for {idp_name} OAuth2 IDP: {oidc_config_url}"
914
1354
  )
915
- ):
1355
+
1356
+ is_localhost = parsed_url.hostname in ("localhost", "127.0.0.1", "::1")
1357
+ if parsed_url.scheme != "https" and not is_localhost:
916
1358
  raise ValueError(
917
- f"An OpenID Connect configuration URL must be set for the {idp_name} OAuth2 IDP "
918
- f"via the {oidc_config_url_env_var} environment variable"
1359
+ f"OIDC configuration URL for {idp_name} OAuth2 IDP "
1360
+ "must use HTTPS (except for localhost)"
919
1361
  )
1362
+
1363
+ # Boolean flags
920
1364
  allow_sign_up = get_env_oauth2_allow_sign_up(idp_name)
921
1365
  auto_login = get_env_oauth2_auto_login(idp_name)
922
- parsed_oidc_config_url = urlparse(oidc_config_url)
923
- is_local_oidc_config_url = parsed_oidc_config_url.hostname in ("localhost", "127.0.0.1")
924
- if parsed_oidc_config_url.scheme != "https" and not is_local_oidc_config_url:
1366
+ use_pkce = _bool_val(f"{idp_prefix}_USE_PKCE", False)
1367
+
1368
+ # Token endpoint auth method validation
1369
+ token_endpoint_auth_method = None
1370
+ if auth_method := _get_optional("TOKEN_ENDPOINT_AUTH_METHOD"):
1371
+ auth_method = auth_method.lower()
1372
+ if auth_method not in _ALLOWED_TOKEN_ENDPOINT_AUTH_METHODS:
1373
+ raise ValueError(
1374
+ f"Invalid TOKEN_ENDPOINT_AUTH_METHOD for {idp_name}. "
1375
+ f"Allowed: {', '.join(sorted(_ALLOWED_TOKEN_ENDPOINT_AUTH_METHODS))}"
1376
+ )
1377
+ token_endpoint_auth_method = auth_method
1378
+
1379
+ # CLIENT_SECRET: required based on TOKEN_ENDPOINT_AUTH_METHOD (OIDC Core §9)
1380
+ client_secret: Optional[str] = None
1381
+
1382
+ # Determine if CLIENT_SECRET is required based on TOKEN_ENDPOINT_AUTH_METHOD:
1383
+ # - "none": CLIENT_SECRET is optional (public clients, RFC 8252 §8.1)
1384
+ # - "client_secret_basic" or "client_secret_post": CLIENT_SECRET is required
1385
+ # - Not set: Default to requiring CLIENT_SECRET (assumes confidential client with
1386
+ # client_secret_basic)
1387
+ #
1388
+ # Note: PKCE (USE_PKCE, RFC 7636) is orthogonal to client authentication. PKCE can be
1389
+ # used with both public clients (no secret) and confidential clients (with secret) to
1390
+ # protect the authorization code from interception.
1391
+
1392
+ if token_endpoint_auth_method == "none":
1393
+ # Public client - no client authentication required
1394
+ client_secret = _get_optional("CLIENT_SECRET")
1395
+ else:
1396
+ # Confidential client (either explicitly set to client_secret_* or using default)
1397
+ # CLIENT_SECRET is required
1398
+ client_secret = _get_required("CLIENT_SECRET", "Client secret")
1399
+
1400
+ # Build scopes: start with required baseline, add custom scopes (deduplicated)
1401
+ scopes = ["openid", "email", "profile"]
1402
+ if custom_scopes := _get_optional("SCOPES"):
1403
+ for scope in custom_scopes.split():
1404
+ if scope and scope not in scopes:
1405
+ scopes.append(scope)
1406
+
1407
+ # Group-based access control
1408
+ groups_attribute_path = _get_optional("GROUPS_ATTRIBUTE_PATH")
1409
+ allowed_groups: list[str] = []
1410
+ if raw_groups := _get_optional("ALLOWED_GROUPS"):
1411
+ # Parse as comma-delimited
1412
+ # Deduplicate while preserving order
1413
+ seen = set()
1414
+ for g in raw_groups.split(","):
1415
+ g = g.strip()
1416
+ if g and g not in seen:
1417
+ allowed_groups.append(g)
1418
+ seen.add(g)
1419
+
1420
+ # Validate: ALLOWED_GROUPS requires GROUPS_ATTRIBUTE_PATH
1421
+ if allowed_groups and not groups_attribute_path:
1422
+ raise ValueError(
1423
+ f"ALLOWED_GROUPS is set for {idp_name} but GROUPS_ATTRIBUTE_PATH is not. "
1424
+ "GROUPS_ATTRIBUTE_PATH must be configured to use group-based access control."
1425
+ )
1426
+
1427
+ # Validate: GROUPS_ATTRIBUTE_PATH requires ALLOWED_GROUPS
1428
+ if groups_attribute_path and not allowed_groups:
925
1429
  raise ValueError(
926
- f"Server metadata URL for {idp_name} OAuth2 IDP "
927
- "must be a valid URL using the https protocol"
1430
+ f"GROUPS_ATTRIBUTE_PATH is set for {idp_name} but ALLOWED_GROUPS is not. "
1431
+ "If you want to extract groups, you must specify which groups are allowed. "
1432
+ "If you don't need group-based access control, remove GROUPS_ATTRIBUTE_PATH."
928
1433
  )
1434
+
1435
+ # Role mapping
1436
+ role_attribute_path = _get_optional("ROLE_ATTRIBUTE_PATH")
1437
+ role_mapping: dict[str, AssignableUserRoleName] = {}
1438
+ if raw_mapping := _get_optional("ROLE_MAPPING"):
1439
+ # Parse role mapping: "IdpRole1:PhoenixRole,IdpRole2:PhoenixRole"
1440
+ for mapping_pair in raw_mapping.split(","):
1441
+ mapping_pair = mapping_pair.strip()
1442
+ if not mapping_pair:
1443
+ continue
1444
+
1445
+ if ":" not in mapping_pair:
1446
+ raise ValueError(
1447
+ f"Invalid ROLE_MAPPING format for {idp_name}: '{mapping_pair}'. "
1448
+ "Expected format: 'IdpRole:PhoenixRole' "
1449
+ "(e.g., 'Owner:ADMIN,Developer:MEMBER')"
1450
+ )
1451
+
1452
+ idp_role, phoenix_role = mapping_pair.split(":", 1)
1453
+ idp_role = idp_role.strip()
1454
+ phoenix_role_upper = phoenix_role.strip().upper()
1455
+
1456
+ if not idp_role:
1457
+ raise ValueError(
1458
+ f"Invalid ROLE_MAPPING for {idp_name}: "
1459
+ f"IDP role cannot be empty in '{mapping_pair}'"
1460
+ )
1461
+
1462
+ # Explicitly reject SYSTEM role (internal-only)
1463
+ if phoenix_role_upper == "SYSTEM":
1464
+ raise ValueError(
1465
+ f"Invalid ROLE_MAPPING for {idp_name}: "
1466
+ f"SYSTEM role cannot be assigned via OAuth2. "
1467
+ f"SYSTEM is an internal-only role for system API keys. "
1468
+ f"Valid roles are: {', '.join(sorted(_VALID_ROLES))}"
1469
+ )
1470
+
1471
+ if phoenix_role_upper not in _VALID_ROLES:
1472
+ valid_roles = ", ".join(sorted(_VALID_ROLES))
1473
+ raise ValueError(
1474
+ f"Invalid ROLE_MAPPING for {idp_name}: "
1475
+ f"'{phoenix_role}' is not a valid Phoenix role. "
1476
+ f"Valid roles are: {valid_roles} (case-insensitive)."
1477
+ )
1478
+
1479
+ role_mapping[idp_role] = phoenix_role_upper # type: ignore[assignment]
1480
+
1481
+ # Get role_attribute_strict setting (defaults to False)
1482
+ role_attribute_strict = _bool_val(f"{idp_prefix}_ROLE_ATTRIBUTE_STRICT", False)
1483
+
1484
+ # Validate role configuration consistency
1485
+ if not role_attribute_path:
1486
+ # If ROLE_ATTRIBUTE_PATH is not configured, other role settings should not be set
1487
+ if role_mapping:
1488
+ raise ValueError(
1489
+ f"Invalid configuration for {idp_name}: ROLE_MAPPING is set but "
1490
+ f"ROLE_ATTRIBUTE_PATH is not configured. ROLE_MAPPING requires "
1491
+ f"ROLE_ATTRIBUTE_PATH to specify where to extract the role from."
1492
+ )
1493
+ if role_attribute_strict:
1494
+ raise ValueError(
1495
+ f"Invalid configuration for {idp_name}: ROLE_ATTRIBUTE_STRICT is set to "
1496
+ f"true but ROLE_ATTRIBUTE_PATH is not configured. ROLE_ATTRIBUTE_STRICT "
1497
+ f"only applies when role extraction is enabled via ROLE_ATTRIBUTE_PATH."
1498
+ )
1499
+
929
1500
  return cls(
930
1501
  idp_name=idp_name,
931
- idp_display_name=getenv(
932
- f"PHOENIX_OAUTH2_{idp_name_upper}_DISPLAY_NAME",
933
- _get_default_idp_display_name(idp_name),
934
- ),
1502
+ idp_display_name=_get_optional("DISPLAY_NAME")
1503
+ or _get_default_idp_display_name(idp_name),
935
1504
  client_id=client_id,
936
1505
  client_secret=client_secret,
937
1506
  oidc_config_url=oidc_config_url,
938
1507
  allow_sign_up=allow_sign_up,
939
1508
  auto_login=auto_login,
1509
+ use_pkce=use_pkce,
1510
+ token_endpoint_auth_method=token_endpoint_auth_method,
1511
+ scopes=" ".join(scopes),
1512
+ groups_attribute_path=groups_attribute_path,
1513
+ allowed_groups=allowed_groups,
1514
+ role_attribute_path=role_attribute_path,
1515
+ role_mapping=role_mapping,
1516
+ role_attribute_strict=role_attribute_strict,
940
1517
  )
941
1518
 
942
1519
 
1520
+ class LDAPGroupRoleMapping(TypedDict):
1521
+ """LDAP group to Phoenix role mapping.
1522
+
1523
+ Attributes:
1524
+ group_dn: LDAP group distinguished name or "*" for wildcard
1525
+ role: Phoenix role name (ADMIN, MEMBER, VIEWER)
1526
+ """
1527
+
1528
+ group_dn: str
1529
+ role: AssignableUserRoleName
1530
+
1531
+
1532
+ def _is_valid_dn(dn: str) -> bool:
1533
+ """Check if a string is a valid LDAP DN syntax."""
1534
+ # Empty DN is valid per RFC 4514 §2 - represents the root DSE (zero RDNs)
1535
+ # https://datatracker.ietf.org/doc/html/rfc4514#section-2
1536
+ if not dn.strip():
1537
+ return True
1538
+ try:
1539
+ parse_dn(dn)
1540
+ return True
1541
+ except LDAPInvalidDnError:
1542
+ return False
1543
+
1544
+
1545
+ @dataclass(frozen=True)
1546
+ class LDAPConfig:
1547
+ """LDAP server configuration for authentication.
1548
+
1549
+ Phoenix uses LDAP (RFC 4510-4519) for user authentication against corporate directories
1550
+ like Active Directory, OpenLDAP, and 389 Directory Server.
1551
+
1552
+ User Identity Strategy
1553
+ ----------------------
1554
+ Phoenix identifies LDAP users using a stable identifier:
1555
+
1556
+ 1. Email (default, recommended for most deployments):
1557
+ When PHOENIX_LDAP_ATTR_UNIQUE_ID is not set, email is used as the identifier.
1558
+ Survives: DN changes, OU moves, renames.
1559
+ If email changes in LDAP: User gets new account (admin can merge manually).
1560
+
1561
+ 2. Unique ID attribute (only if you expect email changes):
1562
+ Set PHOENIX_LDAP_ATTR_UNIQUE_ID to use an immutable LDAP attribute:
1563
+ - Active Directory: "objectGUID"
1564
+ - OpenLDAP: "entryUUID" (RFC 4530)
1565
+ - 389 DS: "nsUniqueId"
1566
+
1567
+ IMPORTANT: Only standard UUID-based attributes are supported. Custom attributes
1568
+ containing 16-character string IDs (e.g., "EMP12345ABCD6789") are NOT supported
1569
+ and will be incorrectly converted. The attribute must contain either:
1570
+ - A 16-byte binary UUID (objectGUID)
1571
+ - A string-format UUID (entryUUID, nsUniqueId)
1572
+
1573
+ Use this only if you expect user emails to change (company rebranding, M&A,
1574
+ frequent name changes). Otherwise, email-based identification is simpler.
1575
+ Survives: Everything including email changes.
1576
+
1577
+ Both modes handle DN changes. The only difference is email change handling.
1578
+ DN is NOT used for identity matching (DNs change too frequently).
1579
+
1580
+ Email as Required Attribute:
1581
+ - Email MUST be present in LDAP for authentication to succeed
1582
+ - Used for Phoenix's user email field (UI, notifications, audit logs)
1583
+ - Provides human-readable identifier for operators
1584
+
1585
+ See: internal_docs/specs/ldap-authentication.md for full design rationale.
1586
+
1587
+ Configuration Pattern
1588
+ ---------------------
1589
+ This class follows the same pattern as OAuth2ClientConfig:
1590
+ - Load from environment variables via from_env()
1591
+ - Validate required fields and format
1592
+ - Provide sensible defaults for optional fields
1593
+ - Document all fields with inline comments
1594
+
1595
+ Attributes
1596
+ ----------
1597
+ Server Connection (RFC 4511):
1598
+ host: LDAP server hostname/IP (required)
1599
+ port: LDAP server port (default: 389 for STARTTLS, 636 for LDAPS)
1600
+ tls_mode: TLS connection mode (default: "starttls")
1601
+ - "starttls": Upgrade from plaintext to TLS on port 389 (recommended)
1602
+ - "ldaps": TLS from connection start on port 636
1603
+ - "none": No encryption (testing only, credentials sent in plaintext)
1604
+ tls_verify: Verify server certificate (default: True, disable only for testing)
1605
+
1606
+ Advanced TLS Configuration (optional, for enterprise deployments):
1607
+ tls_ca_cert_file: Path to custom CA certificate (PEM) for private CAs
1608
+ tls_client_cert_file: Path to client certificate (PEM) for mutual TLS
1609
+ tls_client_key_file: Path to client private key (PEM) for mutual TLS
1610
+
1611
+ Bind Credentials (RFC 4513 §5.1.2 - Simple Authentication):
1612
+ bind_dn: Service account DN for LDAP queries (optional for anonymous bind)
1613
+ bind_password: Service account password (optional for anonymous bind)
1614
+
1615
+ User Search (RFC 4511 §4.5.1):
1616
+ user_search_base_dns: List of base DNs for user searches (searched in order)
1617
+ Example: ["ou=users,dc=example,dc=com"]
1618
+ Multiple: ["ou=employees,dc=corp,dc=com", "ou=contractors,dc=corp,dc=com"]
1619
+ user_search_filter: Filter template with %s placeholder (RFC 4515)
1620
+ Default: "(&(objectClass=user)(sAMAccountName=%s))" (Active Directory)
1621
+ Examples:
1622
+ OpenLDAP: "(&(objectClass=inetOrgPerson)(uid=%s))"
1623
+ 389 DS: "(&(objectClass=person)(uid=%s))"
1624
+
1625
+ Attribute Mapping (RFC 2256, RFC 4524):
1626
+ attr_email: Email attribute name (required, e.g., "mail" or "null")
1627
+ - If set to attribute name: MUST be present in LDAP or login fails
1628
+ - If "null": generates null email marker from unique_id
1629
+ (requires attr_unique_id to be set)
1630
+ attr_display_name: Display name attribute (default: "displayName")
1631
+ - Fallback: Uses email prefix if missing
1632
+ attr_member_of: Group membership attribute (default: "memberOf")
1633
+ - Used when group_search_filter is NOT set
1634
+ - Typical values: "memberOf" (AD/OpenLDAP)
1635
+
1636
+ Group Search (for POSIX/OpenLDAP without memberOf overlay):
1637
+ group_search_filter: Filter template with %s placeholder
1638
+ - When SET: Enables POSIX mode, ignores attr_member_of
1639
+ - When NOT SET: Uses attr_member_of from user entry (AD mode)
1640
+ Example: "(&(objectClass=posixGroup)(memberUid=%s))"
1641
+ group_search_filter_user_attr: User attribute to substitute for %s
1642
+ - When SET: Uses that attribute's value from the user entry (e.g., "uid" → "admin")
1643
+ - When NOT SET: Uses the login username directly (what the user typed at login)
1644
+ For POSIX memberUid filters, the default (login username) is typically correct.
1645
+ group_search_base_dns: List of base DNs for group searches
1646
+ - Required when group_search_filter is set
1647
+ Example: ["ou=groups,dc=example,dc=com"]
1648
+
1649
+ Group to Role Mappings:
1650
+ group_role_mappings: Tuple of dicts mapping LDAP groups to Phoenix roles
1651
+ Format: [{"group_dn": "...", "role": "ADMIN|MEMBER|VIEWER"}]
1652
+ Supports wildcard: {"group_dn": "*", "role": "VIEWER"}
1653
+ Note: Phoenix uses "role" (not "org_role") since it has no organization concept
1654
+
1655
+ IMPORTANT - First Match Wins:
1656
+ Mappings are evaluated in configuration order; the FIRST matching group
1657
+ determines the user's role. This is NOT "highest role wins" - if a user
1658
+ belongs to multiple groups, configuration order (not role hierarchy)
1659
+ determines which role they receive.
1660
+
1661
+ This design matches Grafana's LDAP behavior and common authorization
1662
+ patterns (firewall rules, nginx routing, ACLs). It gives administrators
1663
+ explicit control over precedence.
1664
+
1665
+ Best practice: Order mappings from highest privilege to lowest:
1666
+ [
1667
+ {"group_dn": "cn=admins,...", "role": "ADMIN"},
1668
+ {"group_dn": "cn=developers,...", "role": "MEMBER"},
1669
+ {"group_dn": "*", "role": "VIEWER"}
1670
+ ]
1671
+
1672
+ With this ordering, a user in both "admins" and "developers" groups
1673
+ receives ADMIN (first match). Reversing the order would give MEMBER.
1674
+
1675
+ Sign-Up Control:
1676
+ allow_sign_up: Auto-create users on first login (default: True)
1677
+ True: New users auto-created on first successful LDAP login
1678
+ False: Admins must pre-create users via GraphQL createUser(auth_method: LDAP)
1679
+
1680
+ Examples
1681
+ --------
1682
+ Active Directory:
1683
+ PHOENIX_LDAP_HOST=ldap.corp.example.com
1684
+ PHOENIX_LDAP_PORT=389
1685
+ PHOENIX_LDAP_TLS_MODE=starttls
1686
+ PHOENIX_LDAP_BIND_DN=cn=service,ou=accounts,dc=corp,dc=example,dc=com
1687
+ PHOENIX_LDAP_BIND_PASSWORD=secret
1688
+ PHOENIX_LDAP_USER_SEARCH_BASE_DNS=["ou=users,dc=corp,dc=example,dc=com"]
1689
+ PHOENIX_LDAP_USER_SEARCH_FILTER=(&(objectClass=user)(sAMAccountName=%s))
1690
+ PHOENIX_LDAP_ATTR_EMAIL=mail
1691
+ PHOENIX_LDAP_ATTR_DISPLAY_NAME=displayName
1692
+ PHOENIX_LDAP_ATTR_MEMBER_OF=memberOf
1693
+ PHOENIX_LDAP_GROUP_ROLE_MAPPINGS=[{"group_dn":"cn=admins,ou=groups,dc=corp,dc=example,dc=com","role":"ADMIN"}]
1694
+
1695
+ OpenLDAP with POSIX groups:
1696
+ PHOENIX_LDAP_HOST=ldap.example.com
1697
+ PHOENIX_LDAP_USER_SEARCH_BASE_DNS=["ou=users,dc=example,dc=com"]
1698
+ PHOENIX_LDAP_USER_SEARCH_FILTER=(&(objectClass=inetOrgPerson)(uid=%s))
1699
+ PHOENIX_LDAP_ATTR_EMAIL=mail
1700
+ PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS=["ou=groups,dc=example,dc=com"]
1701
+ PHOENIX_LDAP_GROUP_SEARCH_FILTER=(&(objectClass=posixGroup)(memberUid=%s))
1702
+ PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR=uid
1703
+
1704
+ References
1705
+ ----------
1706
+ - RFC 4510: LDAP Technical Specification Road Map
1707
+ - RFC 4511: LDAP Protocol
1708
+ - RFC 4513: LDAP Authentication & Security
1709
+ - RFC 4515: LDAP Filter String Format
1710
+ - RFC 4524: LDAP mail attribute definition
1711
+ - RFC 2798: inetOrgPerson object class (includes mail)
1712
+ - Grafana LDAP: https://grafana.com/docs/grafana/latest/setup-grafana/configure-access/configure-authentication/ldap/
1713
+ """
1714
+
1715
+ # Server connection (RFC 4511)
1716
+ hosts: tuple[str, ...]
1717
+ port: int = 389
1718
+ tls_mode: Literal["none", "starttls", "ldaps"] = "starttls"
1719
+ tls_verify: bool = True
1720
+
1721
+ # Advanced TLS configuration (optional, for enterprise deployments)
1722
+ tls_ca_cert_file: str | None = None
1723
+ tls_client_cert_file: str | None = None
1724
+ tls_client_key_file: str | None = None
1725
+
1726
+ # Bind credentials (service account, RFC 4513 §5.1.2)
1727
+ bind_dn: str | None = None
1728
+ bind_password: str | None = None
1729
+
1730
+ # User search (RFC 4511 §4.5.1)
1731
+ user_search_base_dns: tuple[str, ...] = ()
1732
+ user_search_filter: str = "(&(objectClass=user)(sAMAccountName=%s))"
1733
+
1734
+ # Attribute mapping (RFC 2798 §9.1.3, §2.3)
1735
+ attr_email: str | None = "mail" # None if explicitly empty (null email marker mode)
1736
+ attr_display_name: str | None = "displayName"
1737
+ attr_member_of: str | None = "memberOf" # Used when group_search_filter is not set
1738
+ attr_unique_id: str | None = None # Optional: objectGUID (AD), entryUUID (OpenLDAP)
1739
+
1740
+ # Group search (for POSIX/OpenLDAP without memberOf)
1741
+ group_search_base_dns: tuple[str, ...] = ()
1742
+ group_search_filter: str | None = None
1743
+ group_search_filter_user_attr: str | None = None # e.g., "uid" for POSIX memberUid
1744
+
1745
+ # Group to role mappings
1746
+ group_role_mappings: tuple[LDAPGroupRoleMapping, ...] = ()
1747
+
1748
+ # Sign-up control
1749
+ allow_sign_up: bool = True
1750
+
1751
+ def __post_init__(self) -> None:
1752
+ if not self.hosts:
1753
+ raise ValueError(f"{ENV_PHOENIX_LDAP_HOST} must contain at least one host")
1754
+
1755
+ @classmethod
1756
+ def from_env(cls) -> "LDAPConfig" | None:
1757
+ """Load LDAP config from environment variables.
1758
+
1759
+ Returns:
1760
+ Optional[LDAPConfig]: LDAP configuration if PHOENIX_LDAP_HOST is set, None otherwise
1761
+
1762
+ Raises:
1763
+ ValueError: If configuration is invalid
1764
+ json.JSONDecodeError: If GROUP_ROLE_MAPPINGS is not valid JSON
1765
+ """
1766
+ host = getenv(ENV_PHOENIX_LDAP_HOST)
1767
+ if not host:
1768
+ return None
1769
+
1770
+ # Import here to avoid circular import at module level
1771
+ from phoenix.server.ldap import canonicalize_dn
1772
+
1773
+ # Normalize and validate host list (remove empty entries from trailing commas, etc.)
1774
+ hosts = tuple(h.strip() for h in host.split(",") if h.strip())
1775
+ if not hosts:
1776
+ raise ValueError(
1777
+ f"{ENV_PHOENIX_LDAP_HOST} must contain at least one non-empty host. "
1778
+ "Example: 'ldap.example.com' or 'dc1.corp.com,dc2.corp.com'"
1779
+ )
1780
+
1781
+ # Parse and validate group role mappings
1782
+ mappings_json = getenv(ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS, "[]")
1783
+ try:
1784
+ group_role_mappings_list = json.loads(mappings_json)
1785
+ except json.JSONDecodeError as e:
1786
+ raise ValueError(
1787
+ f"{ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS} is not valid JSON: {e}. "
1788
+ f"Expected format: [{{'group_dn': '...', 'role': 'ADMIN'}}]"
1789
+ )
1790
+
1791
+ # Validate role mappings structure
1792
+ if not isinstance(group_role_mappings_list, list):
1793
+ raise ValueError(
1794
+ f"{ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS} must be a JSON array. "
1795
+ f"Expected format: [{{'group_dn': '...', 'role': 'ADMIN'}}]"
1796
+ )
1797
+
1798
+ for idx, mapping in enumerate(group_role_mappings_list):
1799
+ if not isinstance(mapping, dict):
1800
+ raise ValueError(
1801
+ f"{ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS}[{idx}] must be an object. "
1802
+ f"Got: {type(mapping).__name__}"
1803
+ )
1804
+ if "group_dn" not in mapping:
1805
+ raise ValueError(
1806
+ f"{ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS}[{idx}] "
1807
+ "missing required field 'group_dn'"
1808
+ )
1809
+ if "role" not in mapping:
1810
+ raise ValueError(
1811
+ f"{ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS}[{idx}] missing required field 'role'"
1812
+ )
1813
+ if not isinstance(mapping["group_dn"], str) or not mapping["group_dn"].strip():
1814
+ raise ValueError(
1815
+ f"{ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS}[{idx}].group_dn "
1816
+ "must be a non-empty string"
1817
+ )
1818
+ # Validate DN syntax and canonicalize (except for wildcard "*")
1819
+ raw_group_dn = mapping["group_dn"].strip()
1820
+ group_dn = canonicalize_dn(raw_group_dn) if raw_group_dn != "*" else raw_group_dn
1821
+ if group_dn is None:
1822
+ raise ValueError(
1823
+ f"{ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS}[{idx}].group_dn "
1824
+ f"has invalid LDAP DN syntax: '{raw_group_dn}'. "
1825
+ f"Expected format: 'cn=GroupName,ou=Groups,dc=example,dc=com'"
1826
+ )
1827
+ # Normalize role to uppercase and validate
1828
+ role_upper = mapping["role"].upper() if isinstance(mapping["role"], str) else ""
1829
+ if role_upper not in _VALID_ROLES:
1830
+ raise ValueError(
1831
+ f"{ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS}[{idx}]: "
1832
+ f"role must be one of {_VALID_ROLES} (case-insensitive). "
1833
+ f"Got: '{mapping['role']}'"
1834
+ )
1835
+ mapping["group_dn"] = group_dn
1836
+ mapping["role"] = role_upper
1837
+
1838
+ # Require at least one role mapping to prevent silent authentication failures
1839
+ # Without mappings, all LDAP users would be denied access with only a debug log,
1840
+ # which is confusing for operators. Fail fast at startup with clear guidance.
1841
+ if not group_role_mappings_list:
1842
+ raise ValueError(
1843
+ f"{ENV_PHOENIX_LDAP_GROUP_ROLE_MAPPINGS} must contain at least one mapping. "
1844
+ f'Example: \'[{{"group_dn": "*", "role": "MEMBER"}}]\' '
1845
+ f"(wildcard '*' grants MEMBER role to all authenticated LDAP users)"
1846
+ )
1847
+
1848
+ # Validate TLS mode
1849
+ tls_mode_str = getenv(ENV_PHOENIX_LDAP_TLS_MODE, "starttls").lower()
1850
+ if tls_mode_str not in ("none", "starttls", "ldaps"):
1851
+ raise ValueError(
1852
+ f"{ENV_PHOENIX_LDAP_TLS_MODE} must be 'none', 'starttls', or 'ldaps'. "
1853
+ f"Got: '{tls_mode_str}'"
1854
+ )
1855
+ tls_mode = cast(Literal["none", "starttls", "ldaps"], tls_mode_str)
1856
+
1857
+ # Parse and validate group_search_base_dns (JSON array of base DNs, optional)
1858
+ attr_member_of = getenv(ENV_PHOENIX_LDAP_ATTR_MEMBER_OF, "memberOf").strip() or None
1859
+ group_search_base_dns_json = getenv(ENV_PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS, "")
1860
+ group_search_filter = getenv(ENV_PHOENIX_LDAP_GROUP_SEARCH_FILTER)
1861
+ group_search_filter_user_attr = getenv(ENV_PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR)
1862
+
1863
+ if group_search_filter and "%s" not in group_search_filter:
1864
+ raise ValueError(
1865
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_FILTER} must contain '%s' placeholder. "
1866
+ f"Got: '{group_search_filter}'"
1867
+ )
1868
+
1869
+ group_search_base_dns_list: list[str] = []
1870
+ if group_search_base_dns_json:
1871
+ try:
1872
+ group_search_base_dns_list = json.loads(group_search_base_dns_json)
1873
+ except json.JSONDecodeError as e:
1874
+ raise ValueError(
1875
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS} is not valid JSON: {e}. "
1876
+ "Expected format: '[\"ou=groups,dc=example,dc=com\"]'"
1877
+ )
1878
+ if not isinstance(group_search_base_dns_list, list):
1879
+ raise ValueError(
1880
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS} must be a JSON array. "
1881
+ "Expected format: '[\"ou=groups,dc=example,dc=com\"]'"
1882
+ )
1883
+ for idx, base_dn in enumerate(group_search_base_dns_list):
1884
+ if not isinstance(base_dn, str):
1885
+ raise ValueError(
1886
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS}[{idx}] must be a string"
1887
+ )
1888
+ stripped = base_dn.strip()
1889
+ if not _is_valid_dn(stripped):
1890
+ raise ValueError(
1891
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS}[{idx}] "
1892
+ f"has invalid LDAP DN syntax: '{base_dn}'. "
1893
+ f"Expected format: 'ou=groups,dc=example,dc=com'"
1894
+ )
1895
+ group_search_base_dns_list[idx] = stripped
1896
+
1897
+ # Validate group search configuration: if filter is set, base DNs are required
1898
+ if group_search_filter and not group_search_base_dns_list:
1899
+ raise ValueError(
1900
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_FILTER} is set but "
1901
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_BASE_DNS} is missing. "
1902
+ f"Both are required for POSIX group search."
1903
+ )
1904
+
1905
+ # Validate group_search_filter_user_attr: only valid when group_search_filter is set
1906
+ if group_search_filter_user_attr and not group_search_filter:
1907
+ raise ValueError(
1908
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR} is set but "
1909
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_FILTER} is not. "
1910
+ f"The user attribute setting only applies to group search filter mode."
1911
+ )
1912
+
1913
+ # Validate attribute name format (no spaces)
1914
+ if group_search_filter_user_attr and " " in group_search_filter_user_attr:
1915
+ suggestion = group_search_filter_user_attr.replace(" ", "")
1916
+ raise ValueError(
1917
+ f"{ENV_PHOENIX_LDAP_GROUP_SEARCH_FILTER_USER_ATTR}="
1918
+ f"'{group_search_filter_user_attr}' contains spaces. "
1919
+ f"LDAP attribute names do not contain spaces. "
1920
+ f"Did you mean '{suggestion}'?"
1921
+ )
1922
+
1923
+ # Security warnings (log, don't fail)
1924
+ tls_verify = _bool_val(ENV_PHOENIX_LDAP_TLS_VERIFY, True)
1925
+ if tls_mode == "none":
1926
+ logger.warning(
1927
+ f"{ENV_PHOENIX_LDAP_TLS_MODE}=none - credentials will be sent in plaintext! "
1928
+ "This is insecure for production."
1929
+ )
1930
+ if tls_mode != "none" and not tls_verify:
1931
+ logger.warning(
1932
+ f"{ENV_PHOENIX_LDAP_TLS_VERIFY} is false - certificates will not be validated! "
1933
+ "This is insecure for production (vulnerable to MITM attacks)."
1934
+ )
1935
+
1936
+ # Parse and validate user_search_base_dns (JSON array of base DNs)
1937
+ user_search_base_dns_json = getenv(ENV_PHOENIX_LDAP_USER_SEARCH_BASE_DNS, "")
1938
+ if not user_search_base_dns_json:
1939
+ raise ValueError(
1940
+ f"{ENV_PHOENIX_LDAP_USER_SEARCH_BASE_DNS} must be set. "
1941
+ "Example: '[\"OU=Users,DC=corp,DC=com\"]'"
1942
+ )
1943
+ user_search_base_dns_list: list[str] = []
1944
+ try:
1945
+ user_search_base_dns_list = json.loads(user_search_base_dns_json)
1946
+ except json.JSONDecodeError as e:
1947
+ raise ValueError(
1948
+ f"{ENV_PHOENIX_LDAP_USER_SEARCH_BASE_DNS} is not valid JSON: {e}. "
1949
+ "Expected format: '[\"OU=Users,DC=corp,DC=com\"]'"
1950
+ )
1951
+ if not isinstance(user_search_base_dns_list, list):
1952
+ raise ValueError(
1953
+ f"{ENV_PHOENIX_LDAP_USER_SEARCH_BASE_DNS} must be a JSON array. "
1954
+ "Expected format: '[\"OU=Users,DC=corp,DC=com\"]'"
1955
+ )
1956
+ if not user_search_base_dns_list:
1957
+ raise ValueError(
1958
+ f"{ENV_PHOENIX_LDAP_USER_SEARCH_BASE_DNS} must contain at least one base DN. "
1959
+ "Example: '[\"OU=Users,DC=corp,DC=com\"]'"
1960
+ )
1961
+ for idx, base_dn in enumerate(user_search_base_dns_list):
1962
+ if not isinstance(base_dn, str):
1963
+ raise ValueError(f"{ENV_PHOENIX_LDAP_USER_SEARCH_BASE_DNS}[{idx}] must be a string")
1964
+ stripped = base_dn.strip()
1965
+ if not _is_valid_dn(stripped):
1966
+ raise ValueError(
1967
+ f"{ENV_PHOENIX_LDAP_USER_SEARCH_BASE_DNS}[{idx}] "
1968
+ f"has invalid LDAP DN syntax: '{base_dn}'. "
1969
+ f"Expected format: 'ou=users,dc=example,dc=com'"
1970
+ )
1971
+ user_search_base_dns_list[idx] = stripped
1972
+
1973
+ # Parse allow_sign_up
1974
+ allow_sign_up = _bool_val(ENV_PHOENIX_LDAP_ALLOW_SIGN_UP, True)
1975
+
1976
+ # Determine default port based on TLS mode (if not explicitly set)
1977
+ # STARTTLS: port 389 (plaintext, then upgrade)
1978
+ # LDAPS: port 636 (TLS from start)
1979
+ default_port = 636 if tls_mode == "ldaps" else 389
1980
+ port = _int_val(ENV_PHOENIX_LDAP_PORT, default_port)
1981
+
1982
+ # Parse advanced TLS configuration (optional)
1983
+ tls_ca_cert_file = getenv(ENV_PHOENIX_LDAP_TLS_CA_CERT_FILE)
1984
+ tls_client_cert_file = getenv(ENV_PHOENIX_LDAP_TLS_CLIENT_CERT_FILE)
1985
+ tls_client_key_file = getenv(ENV_PHOENIX_LDAP_TLS_CLIENT_KEY_FILE)
1986
+
1987
+ # Validate mutual TLS configuration (both cert and key required)
1988
+ if tls_client_cert_file and not tls_client_key_file:
1989
+ raise ValueError(
1990
+ f"{ENV_PHOENIX_LDAP_TLS_CLIENT_CERT_FILE} requires "
1991
+ f"{ENV_PHOENIX_LDAP_TLS_CLIENT_KEY_FILE} to also be set"
1992
+ )
1993
+ if tls_client_key_file and not tls_client_cert_file:
1994
+ raise ValueError(
1995
+ f"{ENV_PHOENIX_LDAP_TLS_CLIENT_KEY_FILE} requires "
1996
+ f"{ENV_PHOENIX_LDAP_TLS_CLIENT_CERT_FILE} to also be set"
1997
+ )
1998
+
1999
+ # Validate file paths exist
2000
+ for env_var, file_path in [
2001
+ (ENV_PHOENIX_LDAP_TLS_CA_CERT_FILE, tls_ca_cert_file),
2002
+ (ENV_PHOENIX_LDAP_TLS_CLIENT_CERT_FILE, tls_client_cert_file),
2003
+ (ENV_PHOENIX_LDAP_TLS_CLIENT_KEY_FILE, tls_client_key_file),
2004
+ ]:
2005
+ if file_path and not os.path.isfile(file_path):
2006
+ raise ValueError(f"{env_var}='{file_path}' does not exist or is not a file")
2007
+
2008
+ # Parse attribute names
2009
+ # attr_email behavior:
2010
+ # - Not set at all → "mail" (backwards compatibility, with deprecation warning)
2011
+ # - "null" sentinel → None (null email marker mode, platform-safe)
2012
+ # - Explicitly empty (PHOENIX_LDAP_ATTR_EMAIL=) → None (null email marker mode)
2013
+ # - Set to value → use that value
2014
+ attr_email_raw = getenv(ENV_PHOENIX_LDAP_ATTR_EMAIL)
2015
+ if attr_email_raw is None:
2016
+ logger.warning(
2017
+ f"{ENV_PHOENIX_LDAP_ATTR_EMAIL} is not set and will be required in the next "
2018
+ f"major version. Set to 'mail' if your LDAP has email, or 'null' if not."
2019
+ )
2020
+ attr_email: str | None = "mail" # Default for backwards compatibility
2021
+ elif attr_email_raw == "" or attr_email_raw.lower() == "null":
2022
+ attr_email = None # Empty or "null" sentinel → null email marker mode
2023
+ else:
2024
+ attr_email = attr_email_raw
2025
+ attr_display_name = (
2026
+ getenv(ENV_PHOENIX_LDAP_ATTR_DISPLAY_NAME, "displayName").strip() or None
2027
+ )
2028
+ attr_unique_id = getenv(ENV_PHOENIX_LDAP_ATTR_UNIQUE_ID, "").strip() or None
2029
+
2030
+ # Validate null email marker mode constraints
2031
+ # When attr_email is "null", we generate markers from unique_id instead
2032
+ if not attr_email:
2033
+ # Constraint 1: unique_id is required for user lookup and marker generation
2034
+ if not attr_unique_id:
2035
+ raise ValueError(
2036
+ f"{ENV_PHOENIX_LDAP_ATTR_UNIQUE_ID} is required when "
2037
+ f"{ENV_PHOENIX_LDAP_ATTR_EMAIL} is 'null'. "
2038
+ f"Without email, unique_id is needed to identify returning users."
2039
+ )
2040
+ # Constraint 2: allow_sign_up must be True (can't pre-provision without unique_id)
2041
+ if not allow_sign_up:
2042
+ raise ValueError(
2043
+ f"{ENV_PHOENIX_LDAP_ALLOW_SIGN_UP} must be True when "
2044
+ f"{ENV_PHOENIX_LDAP_ATTR_EMAIL} is 'null'. "
2045
+ f"Null email markers require auto-provisioning on first login."
2046
+ )
2047
+ # Constraint 3: PHOENIX_ADMINS is not supported (can't compute marker without unique_id)
2048
+ if get_env_admins():
2049
+ raise ValueError(
2050
+ f"PHOENIX_ADMINS is not supported when {ENV_PHOENIX_LDAP_ATTR_EMAIL} "
2051
+ f"is 'null'. Use PHOENIX_LDAP_GROUP_ROLE_MAPPINGS to assign roles."
2052
+ )
2053
+
2054
+ # Validate attribute names don't contain spaces
2055
+ # LDAP attribute names (e.g., objectGUID, entryUUID, mail) never contain spaces.
2056
+ # A space in the config is likely a typo that would cause silent failures.
2057
+ for env_var, attr_value in [
2058
+ (ENV_PHOENIX_LDAP_ATTR_EMAIL, attr_email),
2059
+ (ENV_PHOENIX_LDAP_ATTR_DISPLAY_NAME, attr_display_name),
2060
+ (ENV_PHOENIX_LDAP_ATTR_MEMBER_OF, attr_member_of),
2061
+ (ENV_PHOENIX_LDAP_ATTR_UNIQUE_ID, attr_unique_id),
2062
+ ]:
2063
+ if attr_value and " " in attr_value:
2064
+ suggestion = attr_value.replace(" ", "")
2065
+ raise ValueError(
2066
+ f"{env_var}='{attr_value}' contains spaces. "
2067
+ f"LDAP attribute names do not contain spaces. "
2068
+ f"Did you mean '{suggestion}'?"
2069
+ )
2070
+
2071
+ # Parse and validate search filters
2072
+ user_search_filter = getenv(
2073
+ ENV_PHOENIX_LDAP_USER_SEARCH_FILTER, "(&(objectClass=user)(sAMAccountName=%s))"
2074
+ )
2075
+ if "%s" not in user_search_filter:
2076
+ raise ValueError(
2077
+ f"{ENV_PHOENIX_LDAP_USER_SEARCH_FILTER} must contain '%s' placeholder "
2078
+ f"for username. Got: '{user_search_filter}'"
2079
+ )
2080
+
2081
+ bind_dn = getenv(ENV_PHOENIX_LDAP_BIND_DN, "").strip() or None
2082
+ if bind_dn and not _is_valid_dn(bind_dn):
2083
+ raise ValueError(
2084
+ f"{ENV_PHOENIX_LDAP_BIND_DN} has invalid LDAP DN syntax: '{bind_dn}'. "
2085
+ f"Expected format: 'cn=service,ou=accounts,dc=example,dc=com'"
2086
+ )
2087
+ bind_password = getenv(ENV_PHOENIX_LDAP_BIND_PASSWORD) or None
2088
+ if bind_dn and not bind_password:
2089
+ raise ValueError(
2090
+ f"{ENV_PHOENIX_LDAP_BIND_DN} is set but {ENV_PHOENIX_LDAP_BIND_PASSWORD} is "
2091
+ "missing. Both are required for service account authentication."
2092
+ )
2093
+ if bind_password and not bind_dn:
2094
+ raise ValueError(
2095
+ f"{ENV_PHOENIX_LDAP_BIND_PASSWORD} is set but {ENV_PHOENIX_LDAP_BIND_DN} is "
2096
+ "missing. Both are required for service account authentication."
2097
+ )
2098
+
2099
+ return cls(
2100
+ hosts=hosts,
2101
+ port=port,
2102
+ tls_mode=tls_mode,
2103
+ tls_verify=tls_verify,
2104
+ tls_ca_cert_file=tls_ca_cert_file,
2105
+ tls_client_cert_file=tls_client_cert_file,
2106
+ tls_client_key_file=tls_client_key_file,
2107
+ bind_dn=bind_dn,
2108
+ bind_password=bind_password,
2109
+ user_search_base_dns=tuple(user_search_base_dns_list),
2110
+ user_search_filter=user_search_filter,
2111
+ attr_email=attr_email,
2112
+ attr_display_name=attr_display_name,
2113
+ attr_member_of=attr_member_of,
2114
+ attr_unique_id=attr_unique_id,
2115
+ group_search_base_dns=tuple(group_search_base_dns_list),
2116
+ group_search_filter=group_search_filter,
2117
+ group_search_filter_user_attr=group_search_filter_user_attr,
2118
+ group_role_mappings=tuple(group_role_mappings_list),
2119
+ allow_sign_up=allow_sign_up,
2120
+ )
2121
+
2122
+
2123
+ _OAUTH2_CONFIG_SUFFIXES = (
2124
+ "DISPLAY_NAME", # User-friendly name shown in login UI
2125
+ "CLIENT_ID", # OAuth2 client ID from your identity provider (RFC 6749 §2.2)
2126
+ # OAuth2 client secret (RFC 6749 §2.3.1, required by default, optional with auth method "none")
2127
+ "CLIENT_SECRET",
2128
+ "OIDC_CONFIG_URL", # OpenID Connect discovery URL (.well-known/openid-configuration)
2129
+ "ALLOW_SIGN_UP", # Whether to allow new user registration (default: true)
2130
+ "AUTO_LOGIN", # Automatically redirect to this provider (default: false)
2131
+ "USE_PKCE", # Enable PKCE for authorization code protection (RFC 7636, default: false)
2132
+ "TOKEN_ENDPOINT_AUTH_METHOD", # How to authenticate at token endpoint (OIDC Core §9)
2133
+ # Additional OAuth2 scopes beyond "openid email profile" (RFC 6749 §3.3: space-delimited)
2134
+ "SCOPES",
2135
+ "GROUPS_ATTRIBUTE_PATH", # JMESPath expression to extract groups from ID token
2136
+ "ALLOWED_GROUPS", # Comma-separated list of groups allowed to sign in
2137
+ "ROLE_ATTRIBUTE_PATH", # JMESPath expression to extract role from ID token
2138
+ "ROLE_MAPPING", # Comma-separated list of IDP role to Phoenix role mappings
2139
+ "ROLE_ATTRIBUTE_STRICT", # Whether to deny access if role cannot be extracted/mapped
2140
+ )
2141
+
2142
+
2143
+ _OAUTH2_ENV_VAR_PATTERN = re.compile(
2144
+ rf"^PHOENIX_OAUTH2_(\w+)_({'|'.join(_OAUTH2_CONFIG_SUFFIXES)})$"
2145
+ )
2146
+
2147
+
943
2148
  def get_env_oauth2_settings() -> list[OAuth2ClientConfig]:
944
2149
  """
945
2150
  Retrieves and validates OAuth2/OpenID Connect (OIDC) identity provider configurations from environment variables.
946
2151
 
947
2152
  This function scans the environment for OAuth2 configuration variables and returns a list of
948
- configured identity providers. It supports multiple identity providers simultaneously.
2153
+ configured identity providers. Multiple identity providers can be configured simultaneously,
2154
+ and users will see all enabled providers as login options in the Phoenix UI.
949
2155
 
950
2156
  Environment Variable Pattern:
951
2157
  PHOENIX_OAUTH2_{IDP_NAME}_{CONFIG_TYPE}
952
2158
 
2159
+ Where {IDP_NAME} is any alphanumeric identifier you choose (e.g., GOOGLE, OKTA, KEYCLOAK).
2160
+ The name is case-insensitive and used to group related configuration variables. You can use
2161
+ any name that makes sense for your organization (e.g., COMPANY_SSO, INTERNAL_AUTH).
2162
+
953
2163
  Required Environment Variables for each IDP:
954
2164
  - PHOENIX_OAUTH2_{IDP_NAME}_CLIENT_ID: The OAuth2 client ID issued by the identity provider
955
- - PHOENIX_OAUTH2_{IDP_NAME}_CLIENT_SECRET: The OAuth2 client secret issued by the identity provider
956
- - PHOENIX_OAUTH2_{IDP_NAME}_OIDC_CONFIG_URL: The OpenID Connect configuration URL (must be HTTPS)
2165
+
2166
+ - PHOENIX_OAUTH2_{IDP_NAME}_CLIENT_SECRET: The OAuth2 client secret issued by the identity provider.
2167
+ Required by default for confidential clients. Only optional when TOKEN_ENDPOINT_AUTH_METHOD is
2168
+ explicitly set to "none" (for public clients without client authentication).
2169
+
2170
+ - PHOENIX_OAUTH2_{IDP_NAME}_OIDC_CONFIG_URL: The OpenID Connect configuration URL (must be HTTPS
2171
+ except for localhost). This URL typically ends with /.well-known/openid-configuration and is
2172
+ used to auto-discover OAuth2 endpoints.
957
2173
 
958
2174
  Optional Environment Variables:
959
- - PHOENIX_OAUTH2_{IDP_NAME}_DISPLAY_NAME: A user-friendly name for the identity provider
960
- - PHOENIX_OAUTH2_{IDP_NAME}_ALLOW_SIGN_UP: Whether to allow new user registration (defaults to True)
961
- When set to False, the system will check if the user exists in the database by their email address.
962
- If the user does not exist or has a password set, they will be redirected to the login page with
963
- an error message.
2175
+ - PHOENIX_OAUTH2_{IDP_NAME}_DISPLAY_NAME: A user-friendly name for the identity provider shown in the UI
2176
+
2177
+ - PHOENIX_OAUTH2_{IDP_NAME}_ALLOW_SIGN_UP: Whether to allow new user registration via this OAuth2 provider
2178
+ (defaults to True). When set to False, only existing users can sign in.
2179
+
2180
+ - PHOENIX_OAUTH2_{IDP_NAME}_AUTO_LOGIN: Automatically redirect to this provider's login page, skipping
2181
+ the Phoenix login screen (defaults to False). Useful for single sign-on deployments.
2182
+ Note: Only one provider should have AUTO_LOGIN enabled if you configure multiple IDPs.
2183
+
2184
+ - PHOENIX_OAUTH2_{IDP_NAME}_USE_PKCE: Enable PKCE (Proof Key for Code Exchange) with S256 code challenge
2185
+ method for enhanced security. PKCE protects the authorization code from interception and can be used
2186
+ with both public clients and confidential clients. This setting is orthogonal to client authentication -
2187
+ whether CLIENT_SECRET is required is determined solely by TOKEN_ENDPOINT_AUTH_METHOD, not by USE_PKCE.
2188
+
2189
+ - PHOENIX_OAUTH2_{IDP_NAME}_TOKEN_ENDPOINT_AUTH_METHOD: OAuth2 token endpoint authentication method.
2190
+ This setting determines how the client authenticates with the token endpoint and whether
2191
+ CLIENT_SECRET is required. If not set, defaults to requiring CLIENT_SECRET (confidential client).
2192
+
2193
+ Options:
2194
+ • client_secret_basic: Send credentials in HTTP Basic Auth header (most common).
2195
+ CLIENT_SECRET is required. This is the assumed default behavior if not set.
2196
+ • client_secret_post: Send credentials in POST body (required by some providers).
2197
+ CLIENT_SECRET is required.
2198
+ • none: No client authentication (for public clients).
2199
+ CLIENT_SECRET is not required. Use this for public clients that cannot
2200
+ securely store a client secret, typically in combination with PKCE.
2201
+
2202
+ Most providers work with the default behavior. Set this explicitly only if your provider requires
2203
+ a specific method or if you're configuring a public client.
2204
+
2205
+ - PHOENIX_OAUTH2_{IDP_NAME}_SCOPES: Additional OAuth2 scopes to request (space-separated).
2206
+ These are added to the required baseline scopes "openid email profile". For example, set to
2207
+ "offline_access groups" to request refresh tokens and group information. The baseline scopes
2208
+ are always included and cannot be removed.
2209
+
2210
+ - PHOENIX_OAUTH2_{IDP_NAME}_GROUPS_ATTRIBUTE_PATH: JMESPath expression to extract group/role claims
2211
+ from the OIDC ID token or userinfo endpoint response. See https://jmespath.org for full syntax.
2212
+
2213
+ The path navigates nested JSON structures to find group/role information. This claim is checked
2214
+ from both the ID token and userinfo endpoint (if available). The result is normalized to a list
2215
+ of strings for group matching.
2216
+
2217
+ ⚠️ IMPORTANT: Claim keys with special characters (colons, dots, slashes, hyphens, etc.) MUST be
2218
+ enclosed in double quotes. Examples:
2219
+ • Auth0 namespace: `"https://myapp.com/groups"` (NOT `https://myapp.com/groups`)
2220
+ • AWS Cognito: `"cognito:groups"` (NOT `cognito:groups`)
2221
+ • Keycloak app: `resource_access."my-app".roles` (quotes only around special chars)
2222
+
2223
+ Common JMESPath patterns:
2224
+ • Simple keys: `groups` - extracts top-level array
2225
+ • Nested keys: `resource_access.phoenix.roles` - dot notation for nested objects
2226
+ • Array projection: `teams[*].name` - extracts 'name' field from each object in array
2227
+ • Array indexing: `groups[0]` - gets first element
2228
+
2229
+ Common provider examples:
2230
+ • Google Workspace: `groups`
2231
+ • Azure AD/Entra ID: `roles` or `groups`
2232
+ • Keycloak: `resource_access.phoenix.roles` (nested structure)
2233
+ • AWS Cognito: `"cognito:groups"` (use quotes for colon in key name)
2234
+ • Okta: `groups`
2235
+ • Auth0 (custom namespace): `"https://myapp.com/groups"` (use quotes for special chars)
2236
+ • Custom objects: `teams[*].name` (extract field from array of objects)
2237
+
2238
+ If not set, group-based access control is disabled for this provider.
2239
+
2240
+ - PHOENIX_OAUTH2_{IDP_NAME}_ALLOWED_GROUPS: Comma-separated list of group names that
2241
+ are permitted to sign in. Users must belong to at least one of these groups (extracted via
2242
+ GROUPS_ATTRIBUTE_PATH) to authenticate successfully.
2243
+
2244
+ Example:
2245
+ PHOENIX_OAUTH2_OKTA_ALLOWED_GROUPS="admin,developers,viewers"
2246
+
2247
+ Works together with GROUPS_ATTRIBUTE_PATH to implement group-based access control. If not set,
2248
+ all authenticated users can sign in (subject to ALLOW_SIGN_UP restrictions).
2249
+
2250
+ - PHOENIX_OAUTH2_{IDP_NAME}_ROLE_ATTRIBUTE_PATH: JMESPath expression to extract user role claim
2251
+ from the OIDC ID token or userinfo endpoint response. Similar to GROUPS_ATTRIBUTE_PATH but for
2252
+ extracting a single role value. See https://jmespath.org for full syntax.
2253
+
2254
+ ⚠️ IMPORTANT: Claim keys with special characters MUST be enclosed in double quotes.
2255
+ Examples: `"https://myapp.com/role"`, `"custom:role"`, `user.profile."app-role"`
2256
+
2257
+ Common patterns:
2258
+ • Simple key: `role` - extracts top-level string
2259
+ • Nested key: `user.organization.role` - dot notation for nested objects
2260
+ • Array element: `roles[0]` - gets first role from array
2261
+ • Constant value: `'MEMBER'` - assigns a fixed role to all users from this IDP (no mapping needed)
2262
+ • Conditional logic: `contains(groups[*], 'admin') && 'ADMIN' || 'VIEWER'` - compute role
2263
+ from group membership using logical operators (returns Phoenix role directly, no mapping needed)
2264
+
2265
+ This claim is used with ROLE_MAPPING to automatically assign Phoenix roles (ADMIN, MEMBER, VIEWER)
2266
+ based on the user's role in your identity provider. The extracted role value is matched against
2267
+ keys in ROLE_MAPPING to determine the Phoenix role.
2268
+
2269
+ Advanced: If the JMESPath expression returns a valid Phoenix role name (ADMIN, MEMBER, VIEWER)
2270
+ directly, ROLE_MAPPING is optional - the value will be used as-is after case-insensitive validation.
2271
+
2272
+ ⚠️ Role Update Behavior:
2273
+ • When ROLE_ATTRIBUTE_PATH IS configured: User roles are synchronized from the IDP on EVERY login.
2274
+ This ensures Phoenix roles stay in sync with your IDP's role assignments.
2275
+ • When ROLE_ATTRIBUTE_PATH is NOT configured: User roles are preserved as-is (backward compatibility).
2276
+ New users get VIEWER role (least privilege), existing users keep their current roles.
2277
+
2278
+ - PHOENIX_OAUTH2_{IDP_NAME}_ROLE_MAPPING: Maps identity provider role values to Phoenix roles.
2279
+ Format: "IdpRole1:PhoenixRole1,IdpRole2:PhoenixRole2"
2280
+
2281
+ Phoenix roles (case-insensitive):
2282
+ • ADMIN: Full system access, can manage users and settings
2283
+ • MEMBER: Standard user access, can create and manage own resources
2284
+ • VIEWER: Read-only access, cannot create or modify resources
2285
+
2286
+ Example mappings:
2287
+ PHOENIX_OAUTH2_OKTA_ROLE_MAPPING="Owner:ADMIN,Developer:MEMBER,Guest:VIEWER"
2288
+ PHOENIX_OAUTH2_KEYCLOAK_ROLE_MAPPING="admin:ADMIN,user:MEMBER"
2289
+
2290
+ ⚠️ Security: The SYSTEM role cannot be assigned via OAuth2. Attempts to map to SYSTEM will be rejected.
2291
+
2292
+ Optional Behavior (no mapping required):
2293
+ If ROLE_MAPPING is not configured but ROLE_ATTRIBUTE_PATH is set, the system will use the
2294
+ IDP role value directly if it exactly matches "ADMIN", "MEMBER", or "VIEWER" (case-insensitive).
2295
+ This allows IDPs that already use Phoenix's role names to work without explicit mapping.
2296
+
2297
+ IDP role keys are case-sensitive and must match exactly. Phoenix role values are case-insensitive
2298
+ but will be normalized to uppercase (ADMIN, MEMBER, VIEWER). If a user's IDP role is not in the
2299
+ mapping, behavior depends on ROLE_ATTRIBUTE_STRICT:
2300
+ • strict=false (default): User gets VIEWER role (least privilege)
2301
+ • strict=true: User is denied access
2302
+
2303
+ Works together with ROLE_ATTRIBUTE_PATH. If ROLE_ATTRIBUTE_PATH is set but ROLE_MAPPING is not,
2304
+ the IDP role value is used directly if it matches a valid Phoenix role (ADMIN, MEMBER, VIEWER).
2305
+ If the IDP role doesn't match a valid Phoenix role, behavior depends on ROLE_ATTRIBUTE_STRICT.
2306
+
2307
+ - PHOENIX_OAUTH2_{IDP_NAME}_ROLE_ATTRIBUTE_STRICT: Controls behavior when role cannot be determined
2308
+ from identity provider claims. Defaults to false.
2309
+
2310
+ When true:
2311
+ • Missing role claim → access denied
2312
+ • Role not in ROLE_MAPPING → access denied
2313
+ • Empty/invalid role value → access denied
2314
+
2315
+ When false (default):
2316
+ • Missing/unmapped/invalid role → user gets VIEWER role (least privilege, fail-safe)
2317
+
2318
+ Strict mode is recommended for high-security environments where all users must have explicitly
2319
+ assigned roles. Non-strict mode (default) is more forgiving and suitable for gradual rollout
2320
+ of role mapping.
2321
+
2322
+ Example:
2323
+ PHOENIX_OAUTH2_OKTA_ROLE_ATTRIBUTE_STRICT=true
2324
+
2325
+ Multiple Identity Providers:
2326
+ You can configure multiple IDPs simultaneously. Users will see all configured providers
2327
+ as login options. Each IDP is configured independently with its own set of variables.
2328
+
2329
+ Group-based access control and role mapping are evaluated per-provider:
2330
+ • Groups control access (who can sign in): Users must belong to ALLOWED_GROUPS
2331
+ • Roles control permissions (what users can do): Users are assigned Phoenix roles via ROLE_MAPPING
2332
+ • Groups are checked first, then roles are assigned if access is granted
2333
+ • Each IDP can have different group/role configurations
964
2334
 
965
2335
  Returns:
966
2336
  list[OAuth2ClientConfig]: A list of configured OAuth2 identity providers, sorted alphabetically by IDP name.
@@ -970,20 +2340,82 @@ def get_env_oauth2_settings() -> list[OAuth2ClientConfig]:
970
2340
  ValueError: If required environment variables are missing or invalid.
971
2341
  Specifically, if the OIDC configuration URL is not HTTPS (except for localhost).
972
2342
 
973
- Example:
974
- To configure Google as an identity provider, set these environment variables:
975
- PHOENIX_OAUTH2_GOOGLE_CLIENT_ID=your_client_id
976
- PHOENIX_OAUTH2_GOOGLE_CLIENT_SECRET=your_client_secret
977
- PHOENIX_OAUTH2_GOOGLE_OIDC_CONFIG_URL=https://accounts.google.com/.well-known/openid-configuration
978
- PHOENIX_OAUTH2_GOOGLE_DISPLAY_NAME=Google (optional)
979
- PHOENIX_OAUTH2_GOOGLE_ALLOW_SIGN_UP=true (optional, defaults to true)
2343
+ Examples:
2344
+ Basic configuration with Google:
2345
+ PHOENIX_OAUTH2_GOOGLE_CLIENT_ID=your_client_id
2346
+ PHOENIX_OAUTH2_GOOGLE_CLIENT_SECRET=your_client_secret
2347
+ PHOENIX_OAUTH2_GOOGLE_OIDC_CONFIG_URL=https://accounts.google.com/.well-known/openid-configuration
2348
+
2349
+ With custom display name and auto-login:
2350
+ PHOENIX_OAUTH2_GOOGLE_DISPLAY_NAME=Google Workspace
2351
+ PHOENIX_OAUTH2_GOOGLE_AUTO_LOGIN=true
2352
+
2353
+ With group-based access control (simple path):
2354
+ PHOENIX_OAUTH2_GOOGLE_GROUPS_ATTRIBUTE_PATH=groups
2355
+ PHOENIX_OAUTH2_GOOGLE_ALLOWED_GROUPS=engineering platform-team
2356
+
2357
+ With nested group path (Keycloak):
2358
+ PHOENIX_OAUTH2_KEYCLOAK_GROUPS_ATTRIBUTE_PATH=resource_access.phoenix.roles
2359
+ PHOENIX_OAUTH2_KEYCLOAK_ALLOWED_GROUPS=admin developer
2360
+
2361
+ With special characters in path (AWS Cognito - quotes REQUIRED):
2362
+ PHOENIX_OAUTH2_COGNITO_GROUPS_ATTRIBUTE_PATH='"cognito:groups"'
2363
+ PHOENIX_OAUTH2_COGNITO_ALLOWED_GROUPS=Administrators PowerUsers
2364
+
2365
+ With namespaced claims (Auth0 - quotes REQUIRED):
2366
+ PHOENIX_OAUTH2_AUTH0_GROUPS_ATTRIBUTE_PATH='"https://myapp.com/groups"'
2367
+ PHOENIX_OAUTH2_AUTH0_ALLOWED_GROUPS=admin users
2368
+
2369
+ With array projection (extract names from objects):
2370
+ PHOENIX_OAUTH2_CUSTOM_GROUPS_ATTRIBUTE_PATH=teams[*].name
2371
+ PHOENIX_OAUTH2_CUSTOM_ALLOWED_GROUPS=engineering operations
2372
+
2373
+ With role mapping (simple):
2374
+ PHOENIX_OAUTH2_OKTA_ROLE_ATTRIBUTE_PATH=role
2375
+ PHOENIX_OAUTH2_OKTA_ROLE_MAPPING="Owner:ADMIN,Developer:MEMBER,Viewer:VIEWER"
2376
+
2377
+ With role mapping (nested path for Keycloak):
2378
+ PHOENIX_OAUTH2_KEYCLOAK_ROLE_ATTRIBUTE_PATH=resource_access.phoenix.role
2379
+ PHOENIX_OAUTH2_KEYCLOAK_ROLE_MAPPING="admin:ADMIN,user:MEMBER"
2380
+
2381
+ With role mapping in strict mode (deny unmapped roles):
2382
+ PHOENIX_OAUTH2_OKTA_ROLE_ATTRIBUTE_PATH=role
2383
+ PHOENIX_OAUTH2_OKTA_ROLE_MAPPING="Owner:ADMIN,Developer:MEMBER"
2384
+ PHOENIX_OAUTH2_OKTA_ROLE_ATTRIBUTE_STRICT=true
2385
+
2386
+ With conditional logic to compute role from groups (no mapping needed):
2387
+ PHOENIX_OAUTH2_OKTA_ROLE_ATTRIBUTE_PATH="contains(groups[*], 'admin') && 'ADMIN' || contains(groups[*], 'editor') && 'MEMBER' || 'VIEWER'"
2388
+
2389
+ With both groups and roles (groups control access, roles control permissions):
2390
+ PHOENIX_OAUTH2_OKTA_GROUPS_ATTRIBUTE_PATH=groups
2391
+ PHOENIX_OAUTH2_OKTA_ALLOWED_GROUPS=engineering platform-team
2392
+ PHOENIX_OAUTH2_OKTA_ROLE_ATTRIBUTE_PATH=role
2393
+ PHOENIX_OAUTH2_OKTA_ROLE_MAPPING="Owner:ADMIN,Developer:MEMBER,Guest:VIEWER"
2394
+
2395
+ For public clients using PKCE (no client secret needed):
2396
+ PHOENIX_OAUTH2_MOBILE_CLIENT_ID=mobile_app_id
2397
+ PHOENIX_OAUTH2_MOBILE_OIDC_CONFIG_URL=https://auth.example.com/.well-known/openid-configuration
2398
+ PHOENIX_OAUTH2_MOBILE_TOKEN_ENDPOINT_AUTH_METHOD=none
2399
+ PHOENIX_OAUTH2_MOBILE_USE_PKCE=true
2400
+
2401
+ Multiple identity providers (users can choose):
2402
+ # Google OAuth
2403
+ PHOENIX_OAUTH2_GOOGLE_CLIENT_ID=google_client_id
2404
+ PHOENIX_OAUTH2_GOOGLE_CLIENT_SECRET=google_secret
2405
+ PHOENIX_OAUTH2_GOOGLE_OIDC_CONFIG_URL=https://accounts.google.com/.well-known/openid-configuration
2406
+
2407
+ # Internal Okta
2408
+ PHOENIX_OAUTH2_OKTA_CLIENT_ID=okta_client_id
2409
+ PHOENIX_OAUTH2_OKTA_CLIENT_SECRET=okta_secret
2410
+ PHOENIX_OAUTH2_OKTA_OIDC_CONFIG_URL=https://your-domain.okta.com/.well-known/openid-configuration
2411
+ PHOENIX_OAUTH2_OKTA_GROUPS_ATTRIBUTE_PATH=groups
2412
+ PHOENIX_OAUTH2_OKTA_ALLOWED_GROUPS=engineering
980
2413
  """ # noqa: E501
981
2414
  idp_names = set()
982
- pattern = re.compile(
983
- r"^PHOENIX_OAUTH2_(\w+)_(DISPLAY_NAME|CLIENT_ID|CLIENT_SECRET|OIDC_CONFIG_URL|ALLOW_SIGN_UP|AUTO_LOGIN)$" # noqa: E501
984
- )
985
2415
  for env_var in os.environ:
986
- if (match := pattern.match(env_var)) is not None and (idp_name := match.group(1).lower()):
2416
+ if (match := _OAUTH2_ENV_VAR_PATTERN.match(env_var)) is not None and (
2417
+ idp_name := match.group(1).lower()
2418
+ ):
987
2419
  idp_names.add(idp_name)
988
2420
  return [OAuth2ClientConfig.from_env(idp_name) for idp_name in sorted(idp_names)]
989
2421
 
@@ -1059,26 +2491,45 @@ class DirectoryError(Exception):
1059
2491
 
1060
2492
 
1061
2493
  def get_env_postgres_connection_str() -> Optional[str]:
1062
- pg_user = os.getenv(ENV_PHOENIX_POSTGRES_USER)
1063
- pg_password = os.getenv(ENV_PHOENIX_POSTGRES_PASSWORD)
1064
- pg_host = os.getenv(ENV_PHOENIX_POSTGRES_HOST)
1065
- pg_port = os.getenv(ENV_PHOENIX_POSTGRES_PORT)
1066
- pg_db = os.getenv(ENV_PHOENIX_POSTGRES_DB)
1067
-
1068
- if pg_host and ":" in pg_host:
1069
- pg_host, parsed_port = pg_host.split(":")
1070
- pg_port = pg_port or parsed_port # use the explicitly set port if provided
1071
-
1072
- if pg_host and pg_user and pg_password:
1073
- encoded_password = quote_plus(pg_password)
1074
- connection_str = f"postgresql://{pg_user}:{encoded_password}@{pg_host}"
1075
- if pg_port:
1076
- connection_str = f"{connection_str}:{pg_port}"
1077
- if pg_db:
1078
- connection_str = f"{connection_str}/{pg_db}"
1079
-
1080
- return connection_str
1081
- return None
2494
+ """
2495
+ Build PostgreSQL connection string from environment variables.
2496
+ """
2497
+ pg_host = getenv(ENV_PHOENIX_POSTGRES_HOST, "").rstrip("/")
2498
+ pg_user = getenv(ENV_PHOENIX_POSTGRES_USER)
2499
+ pg_password = getenv(ENV_PHOENIX_POSTGRES_PASSWORD)
2500
+ use_iam_auth = _bool_val(ENV_PHOENIX_POSTGRES_USE_AWS_IAM_AUTH, False)
2501
+
2502
+ if not (pg_host and pg_user):
2503
+ return None
2504
+
2505
+ if use_iam_auth:
2506
+ if pg_password:
2507
+ raise ValueError(
2508
+ f"The environment variable {ENV_PHOENIX_POSTGRES_PASSWORD} is set but will be "
2509
+ "ignored when using AWS RDS IAM authentication "
2510
+ f"({ENV_PHOENIX_POSTGRES_USE_AWS_IAM_AUTH}=true). Authentication tokens will be "
2511
+ "generated using AWS credentials."
2512
+ )
2513
+ connection_str = f"postgresql://{quote(pg_user)}@{pg_host}"
2514
+ else:
2515
+ if not pg_password:
2516
+ raise ValueError(
2517
+ f"The environment variable {ENV_PHOENIX_POSTGRES_PASSWORD} is not set. "
2518
+ "Please set it to the password for the PostgreSQL database."
2519
+ )
2520
+ encoded_user = quote(pg_user)
2521
+ encoded_password = quote(pg_password)
2522
+ connection_str = f"postgresql://{encoded_user}:{encoded_password}@{pg_host}"
2523
+
2524
+ pg_port = getenv(ENV_PHOENIX_POSTGRES_PORT)
2525
+ pg_db = getenv(ENV_PHOENIX_POSTGRES_DB)
2526
+
2527
+ if pg_port:
2528
+ connection_str = f"{connection_str}:{pg_port}"
2529
+ if pg_db:
2530
+ connection_str = f"{connection_str}/{pg_db}"
2531
+
2532
+ return connection_str
1082
2533
 
1083
2534
 
1084
2535
  def _no_local_storage() -> bool:
@@ -1177,7 +2628,7 @@ def ensure_working_dir_if_needed() -> None:
1177
2628
  This is bypassed if a postgres database is configured and a working directory is not set.
1178
2629
  """
1179
2630
  if _no_local_storage():
1180
- pass
2631
+ return
1181
2632
 
1182
2633
  logger.info(f"📋 Ensuring phoenix working directory: {WORKING_DIR}")
1183
2634
  try:
@@ -1271,7 +2722,7 @@ def get_env_host_root_path() -> str:
1271
2722
 
1272
2723
 
1273
2724
  def get_env_collector_endpoint() -> Optional[str]:
1274
- return getenv(ENV_PHOENIX_COLLECTOR_ENDPOINT)
2725
+ return getenv(ENV_PHOENIX_COLLECTOR_ENDPOINT) or getenv(ENV_OTEL_EXPORTER_OTLP_ENDPOINT)
1275
2726
 
1276
2727
 
1277
2728
  def get_env_project_name() -> str:
@@ -1296,7 +2747,36 @@ def get_env_database_schema() -> Optional[str]:
1296
2747
 
1297
2748
 
1298
2749
  def get_env_database_allocated_storage_capacity_gibibytes() -> Optional[float]:
1299
- return _float_val(ENV_PHOENIX_DATABASE_ALLOCATED_STORAGE_CAPACITY_GIBIBYTES)
2750
+ ans = _float_val(ENV_PHOENIX_DATABASE_ALLOCATED_STORAGE_CAPACITY_GIBIBYTES)
2751
+ if ans is not None and ans <= 0:
2752
+ raise ValueError(
2753
+ f"Invalid value for environment variable "
2754
+ f"{ENV_PHOENIX_DATABASE_ALLOCATED_STORAGE_CAPACITY_GIBIBYTES}: "
2755
+ f"{ans}. Value must be a positive number."
2756
+ )
2757
+ return ans
2758
+
2759
+
2760
+ def get_env_database_usage_email_warning_threshold_percentage() -> Optional[float]:
2761
+ ans = _float_val(ENV_PHOENIX_DATABASE_USAGE_EMAIL_WARNING_THRESHOLD_PERCENTAGE)
2762
+ if ans is not None and not (0 <= ans <= 100):
2763
+ raise ValueError(
2764
+ f"Invalid value for environment variable "
2765
+ f"{ENV_PHOENIX_DATABASE_USAGE_EMAIL_WARNING_THRESHOLD_PERCENTAGE}: "
2766
+ f"{ans}. Value must be a percentage between 0 and 100."
2767
+ )
2768
+ return ans
2769
+
2770
+
2771
+ def get_env_database_usage_insertion_blocking_threshold_percentage() -> Optional[float]:
2772
+ ans = _float_val(ENV_PHOENIX_DATABASE_USAGE_INSERTION_BLOCKING_THRESHOLD_PERCENTAGE)
2773
+ if ans is not None and not (0 <= ans <= 100):
2774
+ raise ValueError(
2775
+ f"Invalid value for environment variable "
2776
+ f"{ENV_PHOENIX_DATABASE_USAGE_INSERTION_BLOCKING_THRESHOLD_PERCENTAGE}: "
2777
+ f"{ans}. Value must be a percentage between 0 and 100."
2778
+ )
2779
+ return ans
1300
2780
 
1301
2781
 
1302
2782
  def get_env_enable_prometheus() -> bool:
@@ -1312,6 +2792,30 @@ def get_env_enable_prometheus() -> bool:
1312
2792
  )
1313
2793
 
1314
2794
 
2795
+ def get_env_max_spans_queue_size() -> int:
2796
+ """
2797
+ Gets the maximum spans queue size from the PHOENIX_MAX_SPANS_QUEUE_SIZE environment variable.
2798
+
2799
+ Returns:
2800
+ int: The maximum number of spans to hold in queue before rejecting requests.
2801
+ Defaults to 20,000 if not set.
2802
+
2803
+ Raises:
2804
+ ValueError: If the value is not a positive integer.
2805
+
2806
+ Note:
2807
+ The actual queue size may exceed this limit due to batch processing where a single
2808
+ accepted request can contain multiple spans. This is a heuristic for memory protection.
2809
+ """
2810
+ max_size = _int_val(ENV_PHOENIX_MAX_SPANS_QUEUE_SIZE, 20_000)
2811
+ if max_size <= 0:
2812
+ raise ValueError(
2813
+ f"Invalid value for environment variable {ENV_PHOENIX_MAX_SPANS_QUEUE_SIZE}: "
2814
+ f"{max_size}. Value must be a positive integer."
2815
+ )
2816
+ return max_size
2817
+
2818
+
1315
2819
  def get_env_client_headers() -> dict[str, str]:
1316
2820
  headers = parse_env_headers(getenv(ENV_PHOENIX_CLIENT_HEADERS))
1317
2821
  if (api_key := get_env_phoenix_api_key()) and "authorization" not in [
@@ -1505,6 +3009,10 @@ def get_env_disable_migrations() -> bool:
1505
3009
  return _bool_val(ENV_PHOENIX_DANGEROUSLY_DISABLE_MIGRATIONS, False)
1506
3010
 
1507
3011
 
3012
+ def get_env_mask_internal_server_errors() -> bool:
3013
+ return _bool_val(ENV_PHOENIX_MASK_INTERNAL_SERVER_ERRORS, True)
3014
+
3015
+
1508
3016
  DEFAULT_PROJECT_NAME = "default"
1509
3017
  _KUBERNETES_PHOENIX_PORT_PATTERN = re.compile(r"^tcp://\d{1,3}[.]\d{1,3}[.]\d{1,3}[.]\d{1,3}:\d+$")
1510
3018
 
@@ -1520,11 +3028,92 @@ def get_env_allowed_origins() -> Optional[list[str]]:
1520
3028
  return allowed_origins.split(",")
1521
3029
 
1522
3030
 
3031
+ def get_env_telemetry_enabled() -> bool:
3032
+ """
3033
+ Gets whether telemetry is enabled from the PHOENIX_TELEMETRY_ENABLED environment variable.
3034
+
3035
+ When set to False, disables both FullStory and Scarf.sh tracking regardless of their
3036
+ individual environment variable settings.
3037
+
3038
+ Returns False if external resources are disallowed.
3039
+
3040
+ Returns:
3041
+ bool: True if telemetry is enabled (default), False otherwise.
3042
+ """
3043
+ if not get_env_allow_external_resources():
3044
+ return False
3045
+ return _bool_val(ENV_PHOENIX_TELEMETRY_ENABLED, True)
3046
+
3047
+
3048
+ def get_env_fullstory_org() -> Optional[str]:
3049
+ """
3050
+ Get the FullStory organization ID from environment variables.
3051
+
3052
+ Returns:
3053
+ Optional[str]: The FullStory organization ID if set and telemetry is enabled,
3054
+ None otherwise.
3055
+ """
3056
+ if not get_env_telemetry_enabled():
3057
+ return None
3058
+ return getenv(ENV_PHOENIX_FULLSTORY_ORG)
3059
+
3060
+
3061
+ def get_env_scarf_sh_pixel_id() -> Optional[str]:
3062
+ """
3063
+ Get the Scarf.sh pixel ID from environment variables.
3064
+
3065
+ Returns:
3066
+ Optional[str]: The Scarf.sh pixel ID if set and telemetry is enabled, None otherwise.
3067
+ """
3068
+ if not get_env_telemetry_enabled():
3069
+ return None
3070
+ # Return the phoenix-app-v12 pixel
3071
+ return getenv(ENV_PHOENIX_SCARF_SH_PIXEL_ID) or "98877b05-7d80-493e-ab95-97c104785d1e"
3072
+
3073
+
3074
+ def get_env_management_url() -> Optional[str]:
3075
+ """
3076
+ Gets the value of the PHOENIX_MANAGEMENT_URL environment variable.
3077
+ """
3078
+ return getenv(ENV_PHOENIX_MANAGEMENT_URL)
3079
+
3080
+
3081
+ def get_env_support_email() -> Optional[str]:
3082
+ """
3083
+ Get the support email address from the PHOENIX_SUPPORT_EMAIL environment variable.
3084
+
3085
+ Returns:
3086
+ The support email address if set, None otherwise.
3087
+ """
3088
+ return getenv(ENV_PHOENIX_SUPPORT_EMAIL)
3089
+
3090
+
3091
+ def validate_env_support_email() -> None:
3092
+ """
3093
+ Validate the support email address configured in PHOENIX_SUPPORT_EMAIL.
3094
+
3095
+ Raises:
3096
+ ValueError: If the email address is invalid.
3097
+ """
3098
+ if not (email := get_env_support_email()):
3099
+ return
3100
+ try:
3101
+ validate_email(email, check_deliverability=False)
3102
+ except EmailNotValidError as e:
3103
+ raise ValueError(f"Invalid email in {ENV_PHOENIX_SUPPORT_EMAIL}: '{email}'") from e
3104
+
3105
+
1523
3106
  def verify_server_environment_variables() -> None:
1524
3107
  """Verify that the environment variables are set correctly. Raises an error otherwise."""
1525
3108
  get_env_root_url()
1526
3109
  get_env_phoenix_secret()
1527
3110
  get_env_phoenix_admin_secret()
3111
+ get_env_database_allocated_storage_capacity_gibibytes()
3112
+ get_env_database_usage_email_warning_threshold_percentage()
3113
+ get_env_database_usage_insertion_blocking_threshold_percentage()
3114
+ get_env_max_spans_queue_size()
3115
+ validate_env_support_email()
3116
+ _validate_iam_auth_config()
1528
3117
 
1529
3118
  # Notify users about deprecated environment variables if they are being used.
1530
3119
  if os.getenv("PHOENIX_ENABLE_WEBSOCKETS") is not None:
@@ -1575,3 +3164,86 @@ def _validate_file_exists_and_is_readable(
1575
3164
  f.read(1) # Read just one byte to verify readability
1576
3165
  except Exception as e:
1577
3166
  raise ValueError(f"{description} file is not readable: {e}")
3167
+
3168
+
3169
+ def get_env_allow_external_resources() -> bool:
3170
+ """
3171
+ Gets the value of the PHOENIX_ALLOW_EXTERNAL_RESOURCES environment variable.
3172
+ Defaults to True if not set.
3173
+ """
3174
+ return _bool_val(ENV_PHOENIX_ALLOW_EXTERNAL_RESOURCES, True)
3175
+
3176
+
3177
+ def get_env_postgres_use_iam_auth() -> bool:
3178
+ """
3179
+ Gets whether AWS RDS IAM authentication is enabled for PostgreSQL connections.
3180
+
3181
+ Returns:
3182
+ bool: True if IAM authentication should be used, False otherwise (default)
3183
+ """
3184
+ return _bool_val(ENV_PHOENIX_POSTGRES_USE_AWS_IAM_AUTH, False)
3185
+
3186
+
3187
+ def get_env_postgres_iam_token_lifetime() -> int:
3188
+ """
3189
+ Gets the token lifetime in seconds for AWS RDS IAM authentication pool recycling.
3190
+
3191
+ AWS RDS IAM tokens are valid for 15 minutes (900 seconds). This value should be
3192
+ set slightly lower to ensure connections are recycled before token expiration.
3193
+
3194
+ Returns:
3195
+ int: Token lifetime in seconds (default: 840 = 14 minutes)
3196
+ """
3197
+ lifetime = _int_val(ENV_PHOENIX_POSTGRES_AWS_IAM_TOKEN_LIFETIME_SECONDS, 840)
3198
+ if lifetime <= 0:
3199
+ raise ValueError(
3200
+ f"{ENV_PHOENIX_POSTGRES_AWS_IAM_TOKEN_LIFETIME_SECONDS} must be a positive integer. "
3201
+ f"Got: {lifetime}"
3202
+ )
3203
+ if lifetime > 900:
3204
+ logger.warning(
3205
+ f"{ENV_PHOENIX_POSTGRES_AWS_IAM_TOKEN_LIFETIME_SECONDS} is set to {lifetime} seconds, "
3206
+ f"which exceeds AWS RDS IAM token validity (900 seconds / 15 minutes). "
3207
+ f"Consider setting it to 840 seconds (14 minutes) or less."
3208
+ )
3209
+ return lifetime
3210
+
3211
+
3212
+ def _validate_iam_auth_config() -> None:
3213
+ """
3214
+ Validate AWS RDS IAM authentication configuration if enabled.
3215
+
3216
+ Raises:
3217
+ ImportError: If boto3 is not installed when IAM auth is enabled
3218
+ ValueError: If configuration is invalid
3219
+ """
3220
+ if not get_env_postgres_use_iam_auth():
3221
+ return
3222
+
3223
+ pg_host = getenv(ENV_PHOENIX_POSTGRES_HOST)
3224
+ if not pg_host:
3225
+ return
3226
+
3227
+ try:
3228
+ import boto3 # type: ignore # noqa: F401
3229
+ except ImportError:
3230
+ raise ImportError(
3231
+ f"boto3 is required when {ENV_PHOENIX_POSTGRES_USE_AWS_IAM_AUTH} is enabled. "
3232
+ "Install it with: pip install 'arize-phoenix[aws]'"
3233
+ )
3234
+
3235
+ if not getenv(ENV_PHOENIX_POSTGRES_USER):
3236
+ raise ValueError(
3237
+ f"{ENV_PHOENIX_POSTGRES_USER} must be set when using AWS RDS IAM authentication"
3238
+ )
3239
+
3240
+ try:
3241
+ client = boto3.client("sts") # pyright: ignore
3242
+ client.get_caller_identity() # pyright: ignore
3243
+ logger.info("✓ AWS credentials validated for RDS IAM authentication")
3244
+ except Exception as e:
3245
+ raise ValueError(
3246
+ f"Failed to validate AWS credentials for RDS IAM authentication: {e}. "
3247
+ "Ensure AWS credentials are configured via environment variables, "
3248
+ "~/.aws/credentials, or IAM role."
3249
+ )