[
https://issues.apache.org/jira/browse/SOLR-18280?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18121165#comment-18121165
]
Nick Shanin commented on SOLR-18280:
------------------------------------
🤖 _AI text below_ 🤖
[~epugh]Â I checked the recent 10x nightlies while reviewing this issue:
* [#118|https://ci-builds.apache.org/job/Solr/job/Solr-NightlyTests-10x/118/]
— {{TestReplicationHandler.testUrlAllowList}} passed
* [#119|https://ci-builds.apache.org/job/Solr/job/Solr-NightlyTests-10x/119/]
— build failed before publishing JUnit results
* [#120|https://ci-builds.apache.org/job/Solr/job/Solr-NightlyTests-10x/120/]
— {{TestReplicationHandler.testUrlAllowList}} passed
I have a small local test-only patch that separates the direct allow-list
denial from the leader-to-follower fetch path. Are you still working on this
ticket, or would you like me to finish validating and share the patch?
> reproducible TestReplicationHandler.testUrlAllowList failures
> -------------------------------------------------------------
>
> Key: SOLR-18280
> URL: https://issues.apache.org/jira/browse/SOLR-18280
> Project: Solr
> Issue Type: Test
> Reporter: Chris M. Hostetter
> Assignee: Eric Pugh
> Priority: Major
> Attachments: apache_solr_Solr-NightlyTests-10x_1.log.txt.gz,
> apache_solr_Solr-NightlyTests-10x_2.log.txt.gz,
> apache_solr_Solr-NightlyTests-10x_4.log.txt.gz
>
>
> jenkins stats show a 75% failure rate from
> TestReplicationHandler.testUrlAllowList in the past 7 days – all from 10x
> nightly builds, and all related to seeds that reproduce reliably
> Picking one of the seeds at random, in looking for a "start" point to run bit
> bisect, I found that the failures are not new - these particular seeds _with
> *nightly*_ will fail even if you go several months back.
> Ultimately git bisect identified commit SHA
> {{46479673b836c9364bc3648ceef771fa16b266c6}} (which does not cite a jira) as
> when these failures were introduced.
> Since {{46479673b8}} only modifies test files, it is likely that these
> failures only represent a test bug -- but it's possible they represent a
> "real" bug that our tests only started surfacing after these test changes.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
