connectors_utility 8.6.0.4.pre.20221116T024609Z → 8.6.0.5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/lib/core/connector_job.rb +240 -0
- data/lib/core/connector_settings.rb +14 -7
- data/lib/core/elastic_connector_actions.rb +11 -31
- data/lib/core/scheduler.rb +7 -7
- data/lib/utility/bulk_queue.rb +3 -1
- data/lib/utility/constants.rb +5 -0
- metadata +7 -6
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: bd3d8a98ffaf8965434e1c01467c4ff30d87a1b25e414d5ce36f8a8529178053
|
4
|
+
data.tar.gz: 07d54a470c31ba311aeee4aca41401bca11178d20ed7de72c8dadf7926e910d4
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: eddc02838146053fa61173a5dc6060eb62c92c70930bd5c1338586572af08078288455c65fb2301796d99b338b996ec889966e535bbceaf9f1f39bac6c9888bc
|
7
|
+
data.tar.gz: e0ec181ccc3cb8dd59f123003d873c1218f2e75c6d91569efe075330d913d6bb366c4bcfbc389844b8040a10e22e5b8c797525379d042cb81a2fa2c7e3af08a4
|
@@ -0,0 +1,240 @@
|
|
1
|
+
#
|
2
|
+
# Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
|
3
|
+
# or more contributor license agreements. Licensed under the Elastic License;
|
4
|
+
# you may not use this file except in compliance with the Elastic License.
|
5
|
+
#
|
6
|
+
|
7
|
+
# frozen_string_literal: true
|
8
|
+
|
9
|
+
require 'active_support/core_ext/hash/indifferent_access'
|
10
|
+
require 'connectors/sync_status'
|
11
|
+
require 'core/connector_settings'
|
12
|
+
require 'core/elastic_connector_actions'
|
13
|
+
require 'utility'
|
14
|
+
|
15
|
+
module Core
|
16
|
+
class ConnectorJob
|
17
|
+
DEFAULT_PAGE_SIZE = 100
|
18
|
+
STUCK_THRESHOLD = 60
|
19
|
+
|
20
|
+
def self.fetch_by_id(job_id)
|
21
|
+
es_response = ElasticConnectorActions.get_job(job_id)
|
22
|
+
return nil unless es_response[:found]
|
23
|
+
|
24
|
+
new(es_response)
|
25
|
+
end
|
26
|
+
|
27
|
+
def self.pending_jobs(connectors_ids: [], page_size: DEFAULT_PAGE_SIZE)
|
28
|
+
status_term = { status: Connectors::SyncStatus::PENDING_STATUSES }
|
29
|
+
|
30
|
+
query = { bool: { must: [{ terms: status_term }] } }
|
31
|
+
|
32
|
+
return fetch_jobs_by_query(query, page_size) if connectors_ids.empty?
|
33
|
+
|
34
|
+
query[:bool][:must] << { terms: { 'connector.id' => connectors_ids } }
|
35
|
+
|
36
|
+
fetch_jobs_by_query(query, page_size)
|
37
|
+
end
|
38
|
+
|
39
|
+
def self.orphaned_jobs(page_size = DEFAULT_PAGE_SIZE)
|
40
|
+
connector_ids = ConnectorSettings.fetch_all_connectors.map(&:id)
|
41
|
+
query = { bool: { must_not: { terms: { 'connector.id': connector_ids } } } }
|
42
|
+
fetch_jobs_by_query(query, page_size)
|
43
|
+
end
|
44
|
+
|
45
|
+
def self.delete_jobs(jobs)
|
46
|
+
query = { terms: { '_id': jobs.map(&:id) } }
|
47
|
+
ElasticConnectorActions.delete_jobs_by_query(query)
|
48
|
+
end
|
49
|
+
|
50
|
+
def self.stuck_jobs(connector_id = nil, page_size = DEFAULT_PAGE_SIZE)
|
51
|
+
connector_ids = if connector_id
|
52
|
+
[connector_id]
|
53
|
+
else
|
54
|
+
ConnectorSettings.fetch_native_connectors.map(&:id)
|
55
|
+
end
|
56
|
+
query = {
|
57
|
+
bool: {
|
58
|
+
filter: [
|
59
|
+
{ terms: { 'connector.id': connector_ids } },
|
60
|
+
{ terms: { status: Connectors::SyncStatus::ACTIVE_STATUSES } },
|
61
|
+
{ range: { last_seen: { lte: "now-#{STUCK_THRESHOLD}s" } } }
|
62
|
+
]
|
63
|
+
}
|
64
|
+
}
|
65
|
+
fetch_jobs_by_query(query, page_size)
|
66
|
+
end
|
67
|
+
|
68
|
+
def self.enqueue(_connector_id)
|
69
|
+
nil
|
70
|
+
end
|
71
|
+
|
72
|
+
def id
|
73
|
+
@elasticsearch_response[:_id]
|
74
|
+
end
|
75
|
+
|
76
|
+
def [](property_name)
|
77
|
+
@elasticsearch_response[:_source][property_name]
|
78
|
+
end
|
79
|
+
|
80
|
+
def error
|
81
|
+
self[:error]
|
82
|
+
end
|
83
|
+
|
84
|
+
def status
|
85
|
+
self[:status]
|
86
|
+
end
|
87
|
+
|
88
|
+
def in_progress?
|
89
|
+
status == Connectors::SyncStatus::IN_PROGRESS
|
90
|
+
end
|
91
|
+
|
92
|
+
def canceling?
|
93
|
+
status == Connectors::SyncStatus::CANCELING
|
94
|
+
end
|
95
|
+
|
96
|
+
def suspended?
|
97
|
+
status == Connectors::SyncStatus::SUSPENDED
|
98
|
+
end
|
99
|
+
|
100
|
+
def canceled?
|
101
|
+
status == Connectors::SyncStatus::CANCELED
|
102
|
+
end
|
103
|
+
|
104
|
+
def pending?
|
105
|
+
Connectors::SyncStatus::PENDING_STATUSES.include?(status)
|
106
|
+
end
|
107
|
+
|
108
|
+
def active?
|
109
|
+
Connectors::SyncStatus::ACTIVE_STATUSES.include?(status)
|
110
|
+
end
|
111
|
+
|
112
|
+
def terminated?
|
113
|
+
Connectors::SyncStatus::TERMINAL_STATUSES.include?(status)
|
114
|
+
end
|
115
|
+
|
116
|
+
def connector_snapshot
|
117
|
+
self[:connector] || {}
|
118
|
+
end
|
119
|
+
|
120
|
+
def connector_id
|
121
|
+
connector_snapshot[:id]
|
122
|
+
end
|
123
|
+
|
124
|
+
def index_name
|
125
|
+
connector_snapshot[:index_name]
|
126
|
+
end
|
127
|
+
|
128
|
+
def language
|
129
|
+
connector_snapshot[:language]
|
130
|
+
end
|
131
|
+
|
132
|
+
def service_type
|
133
|
+
connector_snapshot[:service_type]
|
134
|
+
end
|
135
|
+
|
136
|
+
def configuration
|
137
|
+
connector_snapshot[:configuration]
|
138
|
+
end
|
139
|
+
|
140
|
+
def filtering
|
141
|
+
connector_snapshot[:filtering]
|
142
|
+
end
|
143
|
+
|
144
|
+
def pipeline
|
145
|
+
connector_snapshot[:pipeline]
|
146
|
+
end
|
147
|
+
|
148
|
+
def connector
|
149
|
+
@connector ||= ConnectorSettings.fetch_by_id(connector_id)
|
150
|
+
end
|
151
|
+
|
152
|
+
def update_metadata(ingestion_stats = {}, connector_metadata = {})
|
153
|
+
ingestion_stats ||= {}
|
154
|
+
doc = { :last_seen => Time.now }.merge(ingestion_stats)
|
155
|
+
doc[:metadata] = connector_metadata if connector_metadata&.any?
|
156
|
+
ElasticConnectorActions.update_job_fields(id, doc)
|
157
|
+
end
|
158
|
+
|
159
|
+
def done!(ingestion_stats = {}, connector_metadata = {})
|
160
|
+
terminate!(Connectors::SyncStatus::COMPLETED, nil, ingestion_stats, connector_metadata)
|
161
|
+
end
|
162
|
+
|
163
|
+
def error!(message, ingestion_stats = {}, connector_metadata = {})
|
164
|
+
terminate!(Connectors::SyncStatus::ERROR, message, ingestion_stats, connector_metadata)
|
165
|
+
end
|
166
|
+
|
167
|
+
def cancel!(ingestion_stats = {}, connector_metadata = {})
|
168
|
+
terminate!(Connectors::SyncStatus::CANCELED, nil, ingestion_stats, connector_metadata)
|
169
|
+
end
|
170
|
+
|
171
|
+
def with_concurrency_control
|
172
|
+
response = ElasticConnectorActions.get_job(id)
|
173
|
+
|
174
|
+
yield response, response['_seq_no'], response['_primary_term']
|
175
|
+
end
|
176
|
+
|
177
|
+
def make_running!
|
178
|
+
with_concurrency_control do |es_doc, seq_no, primary_term|
|
179
|
+
now = Time.now
|
180
|
+
doc = {
|
181
|
+
status: Connectors::SyncStatus::IN_PROGRESS,
|
182
|
+
started_at: now,
|
183
|
+
last_seen: now,
|
184
|
+
worker_hostname: Socket.gethostname
|
185
|
+
}
|
186
|
+
|
187
|
+
ElasticConnectorActions.update_job_fields(es_doc[:_id], doc, seq_no, primary_term)
|
188
|
+
end
|
189
|
+
end
|
190
|
+
|
191
|
+
def es_source
|
192
|
+
@elasticsearch_response[:_source]
|
193
|
+
end
|
194
|
+
|
195
|
+
private
|
196
|
+
|
197
|
+
def self.fetch_jobs_by_query(query, page_size)
|
198
|
+
results = []
|
199
|
+
offset = 0
|
200
|
+
loop do
|
201
|
+
response = ElasticConnectorActions.search_jobs(query, page_size, offset)
|
202
|
+
|
203
|
+
hits = response.dig('hits', 'hits') || []
|
204
|
+
total = response.dig('hits', 'total', 'value') || 0
|
205
|
+
results += hits.map { |hit| new(hit) }
|
206
|
+
break if results.size >= total
|
207
|
+
offset += hits.size
|
208
|
+
end
|
209
|
+
|
210
|
+
results
|
211
|
+
end
|
212
|
+
|
213
|
+
def initialize(es_response)
|
214
|
+
# TODO: remove the usage of with_indifferent_access. The initialize method should expect a hash argument
|
215
|
+
@elasticsearch_response = es_response.with_indifferent_access
|
216
|
+
end
|
217
|
+
|
218
|
+
def terminate!(status, error = nil, ingestion_stats = {}, connector_metadata = {})
|
219
|
+
ingestion_stats ||= {}
|
220
|
+
ingestion_stats[:total_document_count] = ElasticConnectorActions.document_count(index_name)
|
221
|
+
doc = {
|
222
|
+
:last_seen => Time.now,
|
223
|
+
:completed_at => Time.now,
|
224
|
+
:status => status,
|
225
|
+
:error => error
|
226
|
+
}.merge(ingestion_stats)
|
227
|
+
doc[:canceled_at] = Time.now if status == Connectors::SyncStatus::CANCELED
|
228
|
+
doc[:metadata] = connector_metadata if connector_metadata&.any?
|
229
|
+
ElasticConnectorActions.update_job_fields(id, doc)
|
230
|
+
end
|
231
|
+
|
232
|
+
def seq_no
|
233
|
+
@elasticsearch_response[:_seq_no]
|
234
|
+
end
|
235
|
+
|
236
|
+
def primary_term
|
237
|
+
@elasticsearch_response[:_primary_term]
|
238
|
+
end
|
239
|
+
end
|
240
|
+
end
|
@@ -49,6 +49,11 @@ module Core
|
|
49
49
|
fetch_connectors_by_query(query, page_size)
|
50
50
|
end
|
51
51
|
|
52
|
+
def self.fetch_all_connectors(page_size = DEFAULT_PAGE_SIZE)
|
53
|
+
query = { match_all: {} }
|
54
|
+
fetch_connectors_by_query(query, page_size)
|
55
|
+
end
|
56
|
+
|
52
57
|
def id
|
53
58
|
@elasticsearch_response[:_id]
|
54
59
|
end
|
@@ -130,19 +135,21 @@ module Core
|
|
130
135
|
end
|
131
136
|
|
132
137
|
def update_last_sync!(job)
|
138
|
+
# if job is nil, connector still needs to be updated, to avoid it stuck at in_progress
|
139
|
+
job_status = job&.status || Connectors::SyncStatus::ERROR
|
140
|
+
job_error = job.nil? ? 'Could\'t find the job' : job.error
|
141
|
+
job_error ||= 'unknown error' if job_status == Connectors::SyncStatus::ERROR
|
133
142
|
doc = {
|
134
|
-
:last_sync_status =>
|
143
|
+
:last_sync_status => job_status,
|
135
144
|
:last_synced => Time.now,
|
136
|
-
:last_sync_error =>
|
137
|
-
:error =>
|
145
|
+
:last_sync_error => job_error,
|
146
|
+
:error => job_error
|
138
147
|
}
|
139
|
-
|
140
|
-
if job.terminated?
|
148
|
+
if job&.terminated?
|
141
149
|
doc[:last_indexed_document_count] = job[:indexed_document_count]
|
142
150
|
doc[:last_deleted_document_count] = job[:deleted_document_count]
|
143
151
|
end
|
144
|
-
|
145
|
-
Core::ElasticConnectorActions.update_connector_fields(job.connector_id, doc)
|
152
|
+
Core::ElasticConnectorActions.update_connector_fields(id, doc)
|
146
153
|
end
|
147
154
|
|
148
155
|
private
|
@@ -91,6 +91,17 @@ module Core
|
|
91
91
|
)
|
92
92
|
end
|
93
93
|
|
94
|
+
def delete_jobs_by_query(query)
|
95
|
+
client.delete_by_query(
|
96
|
+
:index => Utility::Constants::JOB_INDEX,
|
97
|
+
:body => { :query => query }
|
98
|
+
)
|
99
|
+
end
|
100
|
+
|
101
|
+
def delete_indices(indices)
|
102
|
+
client.indices.delete(:index => indices, :ignore_unavailable => true)
|
103
|
+
end
|
104
|
+
|
94
105
|
def update_connector_configuration(connector_id, configuration)
|
95
106
|
update_connector_fields(connector_id, :configuration => configuration)
|
96
107
|
end
|
@@ -220,37 +231,6 @@ module Core
|
|
220
231
|
update_connector_fields(connector_id, body)
|
221
232
|
end
|
222
233
|
|
223
|
-
def update_sync(job_id, metadata)
|
224
|
-
body = {
|
225
|
-
:doc => { :last_seen => Time.now }.merge(metadata)
|
226
|
-
}
|
227
|
-
client.update(:index => Utility::Constants::JOB_INDEX, :id => job_id, :body => body)
|
228
|
-
end
|
229
|
-
|
230
|
-
def complete_sync(connector_id, job_id, metadata, error)
|
231
|
-
sync_status = error ? Connectors::SyncStatus::ERROR : Connectors::SyncStatus::COMPLETED
|
232
|
-
|
233
|
-
metadata ||= {}
|
234
|
-
|
235
|
-
update_connector_fields(connector_id,
|
236
|
-
:last_sync_status => sync_status,
|
237
|
-
:last_sync_error => error,
|
238
|
-
:error => error,
|
239
|
-
:last_synced => Time.now,
|
240
|
-
:last_indexed_document_count => metadata[:indexed_document_count],
|
241
|
-
:last_deleted_document_count => metadata[:deleted_document_count])
|
242
|
-
|
243
|
-
body = {
|
244
|
-
:doc => {
|
245
|
-
:status => sync_status,
|
246
|
-
:completed_at => Time.now,
|
247
|
-
:last_seen => Time.now,
|
248
|
-
:error => error
|
249
|
-
}.merge(metadata)
|
250
|
-
}
|
251
|
-
client.update(:index => Utility::Constants::JOB_INDEX, :id => job_id, :body => body)
|
252
|
-
end
|
253
|
-
|
254
234
|
def fetch_document_ids(index_name)
|
255
235
|
page_size = 1000
|
256
236
|
result = []
|
data/lib/core/scheduler.rb
CHANGED
@@ -90,13 +90,6 @@ module Core
|
|
90
90
|
return false
|
91
91
|
end
|
92
92
|
|
93
|
-
# We want to sync when sync never actually happened
|
94
|
-
last_synced = connector_settings[:last_synced]
|
95
|
-
if last_synced.nil? || last_synced.empty?
|
96
|
-
Utility::Logger.info("#{connector_settings.formatted.capitalize} has never synced yet, running initial sync.")
|
97
|
-
return true
|
98
|
-
end
|
99
|
-
|
100
93
|
current_schedule = scheduling_settings[:interval]
|
101
94
|
|
102
95
|
# Don't sync if there is no actual scheduling interval
|
@@ -119,6 +112,13 @@ module Core
|
|
119
112
|
return false
|
120
113
|
end
|
121
114
|
|
115
|
+
# We want to sync when sync never actually happened
|
116
|
+
last_synced = connector_settings[:last_synced]
|
117
|
+
if last_synced.nil? || last_synced.empty?
|
118
|
+
Utility::Logger.info("#{connector_settings.formatted.capitalize} has never synced yet, running initial sync.")
|
119
|
+
return true
|
120
|
+
end
|
121
|
+
|
122
122
|
next_trigger_time = cron_parser.next_time(Time.parse(last_synced))
|
123
123
|
|
124
124
|
# Sync if next trigger for the connector is in past
|
data/lib/utility/bulk_queue.rb
CHANGED
@@ -6,12 +6,14 @@
|
|
6
6
|
|
7
7
|
require 'json'
|
8
8
|
|
9
|
+
require 'utility/constants'
|
10
|
+
|
9
11
|
module Utility
|
10
12
|
class BulkQueue
|
11
13
|
class QueueOverflowError < StandardError; end
|
12
14
|
|
13
15
|
# 500 items or 5MB
|
14
|
-
def initialize(operation_count_threshold =
|
16
|
+
def initialize(operation_count_threshold = Utility::Constants::DEFAULT_MAX_INGESTION_QUEUE_SIZE, size_threshold = Utility::Constants::DEFAULT_MAX_INGESTION_QUEUE_BYTES)
|
15
17
|
@operation_count_threshold = operation_count_threshold.freeze
|
16
18
|
@size_threshold = size_threshold.freeze
|
17
19
|
|
data/lib/utility/constants.rb
CHANGED
@@ -18,5 +18,10 @@ module Utility
|
|
18
18
|
CRAWLER_SERVICE_TYPE = 'elastic-crawler'
|
19
19
|
FILTERING_RULES_FEATURE = 'filtering_rules'
|
20
20
|
FILTERING_ADVANCED_FEATURE = 'filtering_advanced_config'
|
21
|
+
|
22
|
+
# Maximum number of operations in BULK Elasticsearch operation that will ingest the data
|
23
|
+
DEFAULT_MAX_INGESTION_QUEUE_SIZE = 500
|
24
|
+
# Maximum size of either whole BULK Elasticsearch operation or one document in it
|
25
|
+
DEFAULT_MAX_INGESTION_QUEUE_BYTES = 5 * 1024 * 1024
|
21
26
|
end
|
22
27
|
end
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: connectors_utility
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 8.6.0.
|
4
|
+
version: 8.6.0.5
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Elastic
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2022-11-
|
11
|
+
date: 2022-11-21 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: activesupport
|
@@ -106,6 +106,7 @@ files:
|
|
106
106
|
- lib/connectors/crawler/scheduler.rb
|
107
107
|
- lib/connectors/sync_status.rb
|
108
108
|
- lib/connectors_utility.rb
|
109
|
+
- lib/core/connector_job.rb
|
109
110
|
- lib/core/connector_settings.rb
|
110
111
|
- lib/core/elastic_connector_actions.rb
|
111
112
|
- lib/core/filtering/validation_status.rb
|
@@ -130,8 +131,8 @@ homepage: https://github.com/elastic/connectors-ruby
|
|
130
131
|
licenses:
|
131
132
|
- Elastic-2.0
|
132
133
|
metadata:
|
133
|
-
revision:
|
134
|
-
repository: git@github.com:elastic/
|
134
|
+
revision: 39cbb85dbae57a2c92e6e0da272d05aa24ca99a9
|
135
|
+
repository: git@github.com:elastic/connectors-ruby.git
|
135
136
|
post_install_message:
|
136
137
|
rdoc_options: []
|
137
138
|
require_paths:
|
@@ -143,9 +144,9 @@ required_ruby_version: !ruby/object:Gem::Requirement
|
|
143
144
|
version: '0'
|
144
145
|
required_rubygems_version: !ruby/object:Gem::Requirement
|
145
146
|
requirements:
|
146
|
-
- - "
|
147
|
+
- - ">="
|
147
148
|
- !ruby/object:Gem::Version
|
148
|
-
version:
|
149
|
+
version: '0'
|
149
150
|
requirements: []
|
150
151
|
rubygems_version: 3.0.3.1
|
151
152
|
signing_key:
|