mdbkit 0.1.0__tar.gz

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.
mdbkit-0.1.0/LICENSE ADDED
@@ -0,0 +1,21 @@
1
+ MIT License
2
+
3
+ Copyright (c) 2026 Saqib Ameen Subhan
4
+
5
+ Permission is hereby granted, free of charge, to any person obtaining a copy
6
+ of this software and associated documentation files (the "Software"), to deal
7
+ in the Software without restriction, including without limitation the rights
8
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9
+ copies of the Software, and to permit persons to whom the Software is
10
+ furnished to do so, subject to the following conditions:
11
+
12
+ The above copyright notice and this permission notice shall be included in all
13
+ copies or substantial portions of the Software.
14
+
15
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21
+ SOFTWARE.
mdbkit-0.1.0/PKG-INFO ADDED
@@ -0,0 +1,207 @@
1
+ Metadata-Version: 2.4
2
+ Name: mdbkit
3
+ Version: 0.1.0
4
+ Summary: Offline toolkit for MongoDB 4.4+ structured logs: log analysis, slow-query shapes, connection churn, and deterministic index advice. A spiritual successor to mtools' log tools.
5
+ Author: Saqib Ameen Subhan
6
+ License: MIT
7
+ Project-URL: Homepage, https://github.com/saqibameen86/mdbkit
8
+ Project-URL: Issues, https://github.com/saqibameen86/mdbkit/issues
9
+ Keywords: mongodb,logs,logv2,mtools,index,dba,slow-query
10
+ Classifier: Development Status :: 4 - Beta
11
+ Classifier: Environment :: Console
12
+ Classifier: Intended Audience :: System Administrators
13
+ Classifier: Intended Audience :: Developers
14
+ Classifier: License :: OSI Approved :: MIT License
15
+ Classifier: Programming Language :: Python :: 3
16
+ Classifier: Topic :: Database
17
+ Classifier: Topic :: System :: Systems Administration
18
+ Requires-Python: >=3.9
19
+ Description-Content-Type: text/markdown
20
+ License-File: LICENSE
21
+ Provides-Extra: dev
22
+ Requires-Dist: pytest>=7; extra == "dev"
23
+ Dynamic: license-file
24
+
25
+ # mdbkit
26
+
27
+ **An offline toolkit for MongoDB structured logs — log analysis, slow-query shapes, connection churn, and deterministic index advice.**
28
+
29
+ A spiritual successor to [mtools](https://github.com/rueckstiess/mtools)' log tools (`mloginfo`, `mlogfilter`) for the structured JSON log format MongoDB has used since 4.4 — the format mtools never supported. Built for DBAs and ops engineers running self-managed MongoDB 4.4 / 5.0 / 6.0 / 7.0 / 8.0.
30
+
31
+ > **Privacy by design: mdbkit never makes a network call.** It reads log files (or stdin) and writes to stdout. No telemetry, no phoning home, no cloud. Your logs never leave your machine. It is safe to run on air-gapped database hosts.
32
+
33
+ ## Why
34
+
35
+ MongoDB 4.4 switched to structured JSON logging, and the beloved mtools log commands stopped working — the issue has been open since 2020. Meanwhile, index recommendations from MongoDB's Performance Advisor require a paid Atlas tier or Cloud/Ops Manager. If you run Community Edition on your own infrastructure, you're back to reading raw JSON logs with `grep` and `jq`.
36
+
37
+ mdbkit fills that gap: a single, dependency-free CLI that turns structured logs into answers.
38
+
39
+ ## Install
40
+
41
+ ```bash
42
+ pipx install mdbkit # recommended
43
+ # or
44
+ pip install mdbkit
45
+ ```
46
+
47
+ Zero runtime dependencies (pure Python stdlib), so it also installs cleanly on air-gapped hosts from a single wheel:
48
+
49
+ ```bash
50
+ pip download mdbkit -d ./wheels # on a connected machine
51
+ pip install --no-index --find-links ./wheels mdbkit # on the DB host
52
+ ```
53
+
54
+ Requires Python 3.9+.
55
+
56
+ ## Quick start
57
+
58
+ ```bash
59
+ # Overall log summary: versions, restarts, connection counts, error/warning totals
60
+ mdbkit loginfo /var/log/mongodb/mongod.log
61
+
62
+ # Slow queries grouped by query shape (literals stripped), ranked by total time
63
+ mdbkit queries mongod.log
64
+ mdbkit queries mongod.log --sort scanRatio --limit 10 --json
65
+
66
+ # Connection churn by source IP, appName, and driver
67
+ mdbkit connections mongod.log
68
+
69
+ # Filter raw log lines (output stays valid logv2 JSON — chainable)
70
+ mdbkit filter mongod.log --slow 500 --component COMMAND --ns shop.orders
71
+ mdbkit filter mongod.log --from 2026-07-01T08:00:00Z --to 2026-07-01T09:00:00Z | mdbkit queries -
72
+
73
+ # Rotated/compressed logs work directly
74
+ mdbkit queries mongod.log.2.gz
75
+ ```
76
+
77
+ ### Index advice
78
+
79
+ ```bash
80
+ mdbkit advise mongod.log
81
+ ```
82
+
83
+ Produces **candidate** indexes from observed slow-query shapes using the ESR
84
+ (Equality → Sort → Range) guideline, with evidence, confidence, caveats, and a
85
+ validation step for every recommendation:
86
+
87
+ ```
88
+ [1] shop.orders — confidence: HIGH
89
+ query shape : {createdAt:gt, status:eq} sort:{createdAt:-1}
90
+ candidate : { status: 1, createdAt: -1 }
91
+ evidence : COLLSCAN observed in planSummary
92
+ evidence : in-memory sort (hasSortStage) observed
93
+ evidence : examined 251,400 docs to return 73 (3444:1)
94
+ caveat : Every index adds write and storage overhead ...
95
+ validate : Re-run the query with .explain('executionStats') ...
96
+ ```
97
+
98
+ The advice gets sharper if you export your existing indexes and a sampled
99
+ schema. mdbkit never connects to your database — instead it prints small
100
+ `mongosh` scripts you run yourself, so you can read exactly what they do:
101
+
102
+ ```bash
103
+ mdbkit export-script indexes > export_indexes.js
104
+ mdbkit export-script schema > export_schema.js
105
+ mongosh --quiet "mongodb://localhost/shop" export_indexes.js > indexes.json
106
+ mongosh --quiet "mongodb://localhost/shop" export_schema.js > schema.json
107
+
108
+ mdbkit advise mongod.log --indexes indexes.json --schema schema.json
109
+ ```
110
+
111
+ With `--indexes`, mdbkit checks each candidate against your existing indexes
112
+ (flagging when an existing index should already cover the query, or when a
113
+ candidate would make an existing index redundant — it flags, never suggests
114
+ dropping). With `--schema`, it warns about multikey (array) fields,
115
+ low-cardinality fields, and field-name typos. The schema export records field
116
+ **names and types only — no values.**
117
+
118
+ ### Incident triage (beta)
119
+
120
+ ```bash
121
+ mdbkit triage /var/log/mongodb/mongod.log # or a copied log + --no-sysprobe
122
+ ```
123
+
124
+ One command during an incident: election/stepdown timeline, connection
125
+ storms, hot collections, error clusters, slow checkpoints — plus local
126
+ disk/memory/load probes when run on the DB host. Read-only: it never
127
+ connects to the database, and every finding ends with a next step for a
128
+ human to review. Detectors marked beta are validated against real logs
129
+ where available and clearly labeled where broader validation is pending.
130
+
131
+ ### Explain-plan analysis
132
+
133
+ Got a slow query in hand rather than a log? Save its explain output and ask
134
+ mdbkit what's wrong:
135
+
136
+ ```bash
137
+ # in mongosh: EJSON.stringify(db.orders.find({...}).sort({...}).explain("executionStats"))
138
+ mdbkit explain explain.json
139
+ ```
140
+
141
+ You get the plan chain (`SORT -> COLLSCAN`), the examined/returned math, plain-
142
+ English verdicts (full collection scan, blocking in-memory sort, weakly
143
+ selective index, covered query), and — when the plan needs help — the same
144
+ evidence-backed candidate index the advisor would produce. Works with find and
145
+ aggregate explains, classic and SBE (6.0+) plans, and sharded winning plans.
146
+
147
+ ### Design principles
148
+
149
+ * **Deterministic.** Same log in, same advice out. Rules, not AI. Every
150
+ recommendation shows its evidence and its rule of reasoning.
151
+ * **Candidates, not commands.** mdbkit never tells you to blindly run
152
+ `createIndex`, and never advises dropping an index.
153
+ * **Honest about uncertainty.** Shapes seen once are labeled low-confidence;
154
+ `$or`, `$regex`, `$in` and low-selectivity operators carry explicit caveats.
155
+ * **Offline, always.** No network code exists in this codebase.
156
+
157
+ ## What it reads
158
+
159
+ Any MongoDB 4.4+ structured log: `mongod.log`, `mongos` logs, rotated `.gz`
160
+ files, or stdin (`-`). Slow-query lines (`"msg":"Slow query"`) are logged by
161
+ default for operations over `slowms` (100 ms); lower `slowms` or enable
162
+ profiling level 1 to capture more:
163
+
164
+ ```js
165
+ db.setProfilingLevel(1, { slowms: 50 })
166
+ ```
167
+
168
+ Pre-4.4 plain-text logs are detected and politely refused — for those, the
169
+ original mtools still works.
170
+
171
+ ## Roadmap
172
+
173
+ Terminal output is and will remain first-class — this tool is built for the
174
+ Linux box the database actually runs on.
175
+
176
+ * **v0.2** — FTDC (`diagnostic.data`) decoding: offline summaries of the
177
+ metrics MongoDB already records on every node; election/failover timeline
178
+ from REPL events; per-shape drill-down.
179
+ * **v0.3** — shareable Markdown/HTML report export (for tickets and
180
+ post-incident reviews — a convenience layer, never a replacement for the
181
+ terminal).
182
+
183
+ mdbkit is validated against real-world structured logs (tens of thousands of
184
+ lines) in addition to its synthetic test fixtures.
185
+
186
+ ## Bugs, feature requests, questions
187
+
188
+ Please use [GitHub Issues](../../issues) — it keeps problems and fixes public
189
+ so the next person can find them. Real-world log lines that parse wrongly are
190
+ the most valuable bug reports of all (redact literals first!).
191
+
192
+ ## Security
193
+
194
+ mdbkit is offline by design: the codebase contains no network code, never
195
+ executes or evaluates input, and treats every log line as untrusted data
196
+ (strict JSON parsing only — shell constructors are never evaluated). See
197
+ [SECURITY.md](SECURITY.md) for the reporting process.
198
+
199
+ ## Non-affiliation
200
+
201
+ mdbkit is an independent community project. It is **not affiliated with,
202
+ endorsed by, or sponsored by MongoDB, Inc.** "MongoDB" is a registered
203
+ trademark of MongoDB, Inc., used here only to describe compatibility.
204
+
205
+ ## License
206
+
207
+ MIT — see [LICENSE](LICENSE).
mdbkit-0.1.0/README.md ADDED
@@ -0,0 +1,183 @@
1
+ # mdbkit
2
+
3
+ **An offline toolkit for MongoDB structured logs — log analysis, slow-query shapes, connection churn, and deterministic index advice.**
4
+
5
+ A spiritual successor to [mtools](https://github.com/rueckstiess/mtools)' log tools (`mloginfo`, `mlogfilter`) for the structured JSON log format MongoDB has used since 4.4 — the format mtools never supported. Built for DBAs and ops engineers running self-managed MongoDB 4.4 / 5.0 / 6.0 / 7.0 / 8.0.
6
+
7
+ > **Privacy by design: mdbkit never makes a network call.** It reads log files (or stdin) and writes to stdout. No telemetry, no phoning home, no cloud. Your logs never leave your machine. It is safe to run on air-gapped database hosts.
8
+
9
+ ## Why
10
+
11
+ MongoDB 4.4 switched to structured JSON logging, and the beloved mtools log commands stopped working — the issue has been open since 2020. Meanwhile, index recommendations from MongoDB's Performance Advisor require a paid Atlas tier or Cloud/Ops Manager. If you run Community Edition on your own infrastructure, you're back to reading raw JSON logs with `grep` and `jq`.
12
+
13
+ mdbkit fills that gap: a single, dependency-free CLI that turns structured logs into answers.
14
+
15
+ ## Install
16
+
17
+ ```bash
18
+ pipx install mdbkit # recommended
19
+ # or
20
+ pip install mdbkit
21
+ ```
22
+
23
+ Zero runtime dependencies (pure Python stdlib), so it also installs cleanly on air-gapped hosts from a single wheel:
24
+
25
+ ```bash
26
+ pip download mdbkit -d ./wheels # on a connected machine
27
+ pip install --no-index --find-links ./wheels mdbkit # on the DB host
28
+ ```
29
+
30
+ Requires Python 3.9+.
31
+
32
+ ## Quick start
33
+
34
+ ```bash
35
+ # Overall log summary: versions, restarts, connection counts, error/warning totals
36
+ mdbkit loginfo /var/log/mongodb/mongod.log
37
+
38
+ # Slow queries grouped by query shape (literals stripped), ranked by total time
39
+ mdbkit queries mongod.log
40
+ mdbkit queries mongod.log --sort scanRatio --limit 10 --json
41
+
42
+ # Connection churn by source IP, appName, and driver
43
+ mdbkit connections mongod.log
44
+
45
+ # Filter raw log lines (output stays valid logv2 JSON — chainable)
46
+ mdbkit filter mongod.log --slow 500 --component COMMAND --ns shop.orders
47
+ mdbkit filter mongod.log --from 2026-07-01T08:00:00Z --to 2026-07-01T09:00:00Z | mdbkit queries -
48
+
49
+ # Rotated/compressed logs work directly
50
+ mdbkit queries mongod.log.2.gz
51
+ ```
52
+
53
+ ### Index advice
54
+
55
+ ```bash
56
+ mdbkit advise mongod.log
57
+ ```
58
+
59
+ Produces **candidate** indexes from observed slow-query shapes using the ESR
60
+ (Equality → Sort → Range) guideline, with evidence, confidence, caveats, and a
61
+ validation step for every recommendation:
62
+
63
+ ```
64
+ [1] shop.orders — confidence: HIGH
65
+ query shape : {createdAt:gt, status:eq} sort:{createdAt:-1}
66
+ candidate : { status: 1, createdAt: -1 }
67
+ evidence : COLLSCAN observed in planSummary
68
+ evidence : in-memory sort (hasSortStage) observed
69
+ evidence : examined 251,400 docs to return 73 (3444:1)
70
+ caveat : Every index adds write and storage overhead ...
71
+ validate : Re-run the query with .explain('executionStats') ...
72
+ ```
73
+
74
+ The advice gets sharper if you export your existing indexes and a sampled
75
+ schema. mdbkit never connects to your database — instead it prints small
76
+ `mongosh` scripts you run yourself, so you can read exactly what they do:
77
+
78
+ ```bash
79
+ mdbkit export-script indexes > export_indexes.js
80
+ mdbkit export-script schema > export_schema.js
81
+ mongosh --quiet "mongodb://localhost/shop" export_indexes.js > indexes.json
82
+ mongosh --quiet "mongodb://localhost/shop" export_schema.js > schema.json
83
+
84
+ mdbkit advise mongod.log --indexes indexes.json --schema schema.json
85
+ ```
86
+
87
+ With `--indexes`, mdbkit checks each candidate against your existing indexes
88
+ (flagging when an existing index should already cover the query, or when a
89
+ candidate would make an existing index redundant — it flags, never suggests
90
+ dropping). With `--schema`, it warns about multikey (array) fields,
91
+ low-cardinality fields, and field-name typos. The schema export records field
92
+ **names and types only — no values.**
93
+
94
+ ### Incident triage (beta)
95
+
96
+ ```bash
97
+ mdbkit triage /var/log/mongodb/mongod.log # or a copied log + --no-sysprobe
98
+ ```
99
+
100
+ One command during an incident: election/stepdown timeline, connection
101
+ storms, hot collections, error clusters, slow checkpoints — plus local
102
+ disk/memory/load probes when run on the DB host. Read-only: it never
103
+ connects to the database, and every finding ends with a next step for a
104
+ human to review. Detectors marked beta are validated against real logs
105
+ where available and clearly labeled where broader validation is pending.
106
+
107
+ ### Explain-plan analysis
108
+
109
+ Got a slow query in hand rather than a log? Save its explain output and ask
110
+ mdbkit what's wrong:
111
+
112
+ ```bash
113
+ # in mongosh: EJSON.stringify(db.orders.find({...}).sort({...}).explain("executionStats"))
114
+ mdbkit explain explain.json
115
+ ```
116
+
117
+ You get the plan chain (`SORT -> COLLSCAN`), the examined/returned math, plain-
118
+ English verdicts (full collection scan, blocking in-memory sort, weakly
119
+ selective index, covered query), and — when the plan needs help — the same
120
+ evidence-backed candidate index the advisor would produce. Works with find and
121
+ aggregate explains, classic and SBE (6.0+) plans, and sharded winning plans.
122
+
123
+ ### Design principles
124
+
125
+ * **Deterministic.** Same log in, same advice out. Rules, not AI. Every
126
+ recommendation shows its evidence and its rule of reasoning.
127
+ * **Candidates, not commands.** mdbkit never tells you to blindly run
128
+ `createIndex`, and never advises dropping an index.
129
+ * **Honest about uncertainty.** Shapes seen once are labeled low-confidence;
130
+ `$or`, `$regex`, `$in` and low-selectivity operators carry explicit caveats.
131
+ * **Offline, always.** No network code exists in this codebase.
132
+
133
+ ## What it reads
134
+
135
+ Any MongoDB 4.4+ structured log: `mongod.log`, `mongos` logs, rotated `.gz`
136
+ files, or stdin (`-`). Slow-query lines (`"msg":"Slow query"`) are logged by
137
+ default for operations over `slowms` (100 ms); lower `slowms` or enable
138
+ profiling level 1 to capture more:
139
+
140
+ ```js
141
+ db.setProfilingLevel(1, { slowms: 50 })
142
+ ```
143
+
144
+ Pre-4.4 plain-text logs are detected and politely refused — for those, the
145
+ original mtools still works.
146
+
147
+ ## Roadmap
148
+
149
+ Terminal output is and will remain first-class — this tool is built for the
150
+ Linux box the database actually runs on.
151
+
152
+ * **v0.2** — FTDC (`diagnostic.data`) decoding: offline summaries of the
153
+ metrics MongoDB already records on every node; election/failover timeline
154
+ from REPL events; per-shape drill-down.
155
+ * **v0.3** — shareable Markdown/HTML report export (for tickets and
156
+ post-incident reviews — a convenience layer, never a replacement for the
157
+ terminal).
158
+
159
+ mdbkit is validated against real-world structured logs (tens of thousands of
160
+ lines) in addition to its synthetic test fixtures.
161
+
162
+ ## Bugs, feature requests, questions
163
+
164
+ Please use [GitHub Issues](../../issues) — it keeps problems and fixes public
165
+ so the next person can find them. Real-world log lines that parse wrongly are
166
+ the most valuable bug reports of all (redact literals first!).
167
+
168
+ ## Security
169
+
170
+ mdbkit is offline by design: the codebase contains no network code, never
171
+ executes or evaluates input, and treats every log line as untrusted data
172
+ (strict JSON parsing only — shell constructors are never evaluated). See
173
+ [SECURITY.md](SECURITY.md) for the reporting process.
174
+
175
+ ## Non-affiliation
176
+
177
+ mdbkit is an independent community project. It is **not affiliated with,
178
+ endorsed by, or sponsored by MongoDB, Inc.** "MongoDB" is a registered
179
+ trademark of MongoDB, Inc., used here only to describe compatibility.
180
+
181
+ ## License
182
+
183
+ MIT — see [LICENSE](LICENSE).
@@ -0,0 +1,3 @@
1
+ """mdbkit — an offline toolkit for MongoDB structured logs and index advice."""
2
+
3
+ __version__ = "0.1.0"