You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Maxim Muzafarov (Jira)" <ji...@apache.org> on 2024/10/01 11:45:00 UTC
[jira] [Comment Edited] (CASSANDRA-19365) invalid EstimatedHistogramReservoirSnapshot::getValue values due to race condition in DecayingEstimatedHistogramReservoir
[ https://issues.apache.org/jira/browse/CASSANDRA-19365?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17886132#comment-17886132 ]
Maxim Muzafarov edited comment on CASSANDRA-19365 at 10/1/24 11:44 AM:
-----------------------------------------------------------------------
Yes, the [trunk|https://github.com/apache/cassandra/pull/3543/files], [5.0|https://github.com/apache/cassandra/pull/3587/files], and [4.1|https://github.com/apache/cassandra/pull/3588/files] patches are identical. The CI is here and looks good as well.
[https://ci-cassandra.apache.org/view/patches/job/Cassandra-devbranch-5/74/]
[https://ci-cassandra.apache.org/view/patches/job/Cassandra-devbranch-5/73/]
[https://ci-cassandra.apache.org/job/Cassandra-devbranch-before-5-test/]
For 4.0 I think it would be better not to cherry-pick the patch to this branch without having a strong need (or direct request from users) and for the sake of stability, as several commits [8960b4d|https://github.com/apache/cassandra/commit/d2923275e360a1ee9db498e748c269f701bb3a8b], [910bbb9|https://github.com/apache/cassandra/commit/af90c3634007ba9b2967267d6f8383d38cdff85f] between 4.0 -> 4.1 affect the initial patch, so from an implementation standpoint it's a slightly different patch for 4.0. The risks of maintaining a different implementation outweigh the benefits of fixing the problem, from my point of view.
was (Author: mmuzaf):
Yes, the [trunk|https://github.com/apache/cassandra/pull/3543/files], [5.0|https://github.com/apache/cassandra/pull/3587/files], and [4.1|https://github.com/apache/cassandra/pull/3588/files] patches are identical. The CI is here and looks good as well.
[https://ci-cassandra.apache.org/view/patches/job/Cassandra-devbranch-5/74/]
[https://ci-cassandra.apache.org/view/patches/job/Cassandra-devbranch-5/73/]
[https://ci-cassandra.apache.org/job/Cassandra-devbranch-before-5-test/]
For 4.0 I think it would be better not to cherry-pick the patch to this branch without having a strong need (or direct request from users) and for the sake of stability, as several commits between 4.0 -> 4.1 affect the patch [8960b4d|https://github.com/apache/cassandra/commit/d2923275e360a1ee9db498e748c269f701bb3a8b], [910bbb9|https://github.com/apache/cassandra/commit/af90c3634007ba9b2967267d6f8383d38cdff85f], so from an implementation standpoint it's a slightly different patch. The risks of maintaining a different implementation outweigh the benefits of fixing the problem, from my point of view.
> invalid EstimatedHistogramReservoirSnapshot::getValue values due to race condition in DecayingEstimatedHistogramReservoir
> -------------------------------------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-19365
> URL: https://issues.apache.org/jira/browse/CASSANDRA-19365
> Project: Cassandra
> Issue Type: Bug
> Components: Observability/Metrics
> Reporter: Jakub Zytka
> Assignee: Maxim Muzafarov
> Priority: Normal
> Fix For: 4.1.x, 5.0.x, 5.x
>
> Attachments: ci_summary.html, result_details.tar.gz
>
> Time Spent: 4h 20m
> Remaining Estimate: 0h
>
> `DecayingEstimatedHistogramReservoir` has a race condition between `update` and `rescaleIfNeeded`.
> A sample which ends up (`update`) in an already scaled decayingBucket (`rescaleIfNeeded`) may still use a non-scaled weight because `decayLandmark` has not been updated yet at the moment of `update`.
>
> The observed consequence was flooding of the cluster with speculative retries (we happened to hit low-percentile buckets with overweight samples, which drove p99 below true p50 for a long time).
> Please note that despite the manifestation being similar to CASSANDRA-19330, these are two distinct bugs in their own right.
> This bug affects versions 4.0+
> On 3.11 there's locking in DEHR. I did not check earlier versions.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@cassandra.apache.org
For additional commands, e-mail: commits-help@cassandra.apache.org