[
https://issues.apache.org/jira/browse/CASSANDRA-19428?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17824215#comment-17824215
]
Ekaterina Dimitrova edited comment on CASSANDRA-19428 at 3/6/24 11:26 PM:
--------------------------------------------------------------------------
I created some trunk [patch|https://github.com/apache/cassandra/pull/3162]
based on a few works from Jonathan Ellis, Michael Marchall, and a PR in
progress by [~pkolaczk](with the idea to help him make rebase easier), but now
CI fails consistently for one test which times out:
testFailedRangeIteratorOnMultiIndexesQuery -
[https://app.circleci.com/pipelines/github/ekaterinadimitrova2/cassandra/2666/workflows/fdd384e4-931b-408c-8496-dc22e4757ec4/jobs/58263/tests]
I can reproduce it locally. I haven't investigated it yet, but I have a few
immediate observations so far:
- despite that, the test is testFailedRangeIteratorOn*Multi*IndexesQuery, with
stress on Multi, I see only one index created. ([~mikea] ? You are the author
of this test, so checking here if you have any comments)
- The moment we disable the injection, we start timing out after that. If I
swap to check first the assertions that do not throw exceptions and then the
ones that will throw after the injected failure - the test succeeds. I will
have to check that
I am not sure whether I will be able to get back to this before next week and
how much time I will have, but I wanted to show what I am doing and check
whether it makes sense so far.
CC [~maedhroz]
was (Author: e.dimitrova):
I created some trunk [patch|https://github.com/apache/cassandra/pull/3162]
based on a few works from Jonathan Ellis, Michael Marchall, and a PR in
progress by [~pkolaczk](with the idea to help him make rebase easier), but now
CI fails consistently for one test which times out:
testFailedRangeIteratorOnMultiIndexesQuery -
[https://app.circleci.com/pipelines/github/ekaterinadimitrova2/cassandra/2666/workflows/fdd384e4-931b-408c-8496-dc22e4757ec4/jobs/58263/tests]
I can reproduce it locally. I haven't investigated it yet, but I have a few
immediate observations so far:
- despite that, the test is testFailedRangeIteratorOn*Multi*IndexesQuery, with
stress on Multi, I see only one index created. ([~mikea] ? You are the author
of this test, so checking here if you have any comments)
- The moment we disable the injection, we start timing out after that. If I
swap to check first the assertions that do not throw exceptions and then the
ones that will throw after the injected failure - the test succeeds. I will
have to check that
I am not sure whether I will be able to get back to this before next week and
how much time I will have, but I wanted to show what I am doing and check
whether it makes sense so far.
> Clean up KeyRangeItearor classes
> --------------------------------
>
> Key: CASSANDRA-19428
> URL: https://issues.apache.org/jira/browse/CASSANDRA-19428
> Project: Cassandra
> Issue Type: Improvement
> Components: Feature/2i Index
> Reporter: Ekaterina Dimitrova
> Assignee: Ekaterina Dimitrova
> Priority: Low
> Fix For: 5.x
>
>
> Remove KeyRangeIterator.current and simplify
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
