[
https://issues.apache.org/jira/browse/ATLAS-4882?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17859550#comment-17859550
]
ASF subversion and git services commented on ATLAS-4882:
--------------------------------------------------------
Commit e9a36e75bc220d7957dd0c1963d74e25d002812d in atlas's branch
refs/heads/master from Pinal Shah
[ https://gitbox.apache.org/repos/asf?p=atlas.git;h=e9a36e75b ]
ATLAS-4882: Export/Import: Export exits with 'Found 0 entities'
Signed-off-by: Pinal Shah <pinal.s...@freestoneinfotech.com>
> Export/Import: Export exits with "Found 0 entities"
> ----------------------------------------------------
>
> Key: ATLAS-4882
> URL: https://issues.apache.org/jira/browse/ATLAS-4882
> Project: Atlas
> Issue Type: Bug
> Components: atlas-core
> Reporter: Pinal Shah
> Assignee: Pinal Shah
> Priority: Major
>
> *Issue:*
> Export during ingestion fails giving Found 0 entities in the logs
> Ingestion meaning Atlas is consuming messages
> *Steps to Repro:*
> * Make sure backend has above 1M entities
> * Start creating tables under `db1@cm`
> * Start export for `db1@cm`
> *
> {code:java}
> curl -v -X POST -u admin:admin -H "Content-Type: application/json"
> "http://<>/api/atlas/admin/export" -d
> '{"itemsToExport":[{"typeName":"hive_db","uniqueAttributes": {
> "qualifiedName": "db1@cm"
> }}],"options":{"fetchType":"full","replicatedTo":"cm"}} > export1.zip{code}
> * It fails after sometime.
> *When is the issue seen?*
> It occurs when there is huge amount of data in backend and Atlas is consuming
> messages linked to entity of which export is running
> *Analysis to find Root cause:*
> * when there is huge amount of data in backend, export FAILS
> * when there is huge amount of data in backend but less tables under it,
> then also export FAILS
> * if background consumption stops, export PASS
> * if consumption is of different entities then requested in export, export
> PASS
> * export query to find starting object uses below query, where has clause to
> check property is expensive
> {code:java}
> g.V().has('_typeName','hive_db').has('Referenceable.qualifiedName','db6@cm').has('__guid').values('__guid'){code}
> - has('__guid') queries solr [(35x_t <> null)]:vertex_index
> - below is the timetaken in the solr logs
> {code:java}
> 2024-06-14 02:38:56.218 INFO (qtp1158676965-19) [c:vertex_index s:shard1
> r:core_node2 x:vertex_index_shard1_replica_n1] o.a.s.c.S.Request
> [vertex_index_shard1_replica_n1] webapp=/solr path=/select
> params={q=:&stateVer=vertex_index:12&fl=id&start=0&fq=35x_t:*+&rows=500000&wt=javabin&version=2}
> hits=1681928 status=0 QTime=4227
> 2024-06-14 02:40:23.945 INFO (qtp1158676965-16) [c:vertex_index s:shard1
> r:core_node2 x:vertex_index_shard1_replica_n1] o.a.s.c.S.Request
> [vertex_index_shard1_replica_n1] webapp=/solr path=/select
> params={q=:&stateVer=vertex_index:12&fl=id&start=500000&fq=35x_t:*+&rows=500000&wt=javabin&version=2}
> hits=1682086 status=0 QTime=787
> 2024-06-14 02:41:37.703 INFO (qtp1158676965-14) [c:vertex_index s:shard1
> r:core_node2 x:vertex_index_shard1_replica_n1] o.a.s.c.S.Request
> [vertex_index_shard1_replica_n1] webapp=/solr path=/select
> params={q=:&stateVer=vertex_index:12&fl=id&start=1000000&fq=35x_t:*+&rows=500000&wt=javabin&version=2}
> hits=1682216 status=0 QTime=1962
> 2024-06-14 02:42:20.715 INFO (qtp1158676965-20) [c:vertex_index s:shard1
> r:core_node2 x:vertex_index_shard1_replica_n1] o.a.s.c.S.Request
> [vertex_index_shard1_replica_n1] webapp=/solr path=/select
> params={q=:&stateVer=vertex_index:12&fl=id&start=1500000&fq=35x_t:*+&rows=500000&wt=javabin&version=2}
> hits=1682363 status=0 QTime=4465 {code}
> - ran same query through gremlin shell while ingestion is happening it
> doesn't fail
> - time taken for above gremlin query in code when ingestion
> : 214825ms
> - time takem for above gremlin query in gremlin shell when ingestion :
> 104641ms
> - time taken for above gremlin query when no ingestion
> : 181682ms
> Still Root cause is unknown
> *WorkAround:*
> - Remove .has('__guid') clause from below, it is very quick and issue is not
> reproducible.
> {code:java}
> g.V().has('_typeName','hive_db').has('Referenceable.qualifiedName','db6@cm').has('__guid').values('__guid'){code}
> *Tests:*
> * upgrded tinkerpop and janusgraph version but didn't help
> * invalid property doesn't throw any exception or not existence of property
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
