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 2023/03/01 12:43:00 UTC
[jira] [Comment Edited] (CASSANDRA-18131) LongBTreeTest times out after btree improvements from CASSANDRA-15510
[ https://issues.apache.org/jira/browse/CASSANDRA-18131?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17694782#comment-17694782 ]
Maxim Muzafarov edited comment on CASSANDRA-18131 at 3/1/23 12:42 PM:
----------------------------------------------------------------------
[~mck], [~benedict],
The reason the test suite hangs is that the number of running tests configured by the {{perThreadTrees}} configuration parameter is too large to complete within a reasonable time (the test suite timeout is 90 minutes). There is no such problem for the {{cassandra-3.11}} branch, as {{perThreadTrees}} is set to 100, so the test completes within 15 minutes. See [LongBTreeTest.java#L61|https://github.com/apache/cassandra/blob/cassandra-3.11/test/burn/org/apache/cassandra/utils/LongBTreeTest.java#L61].
The total number of "runs" has been changed several times to fit the timeout, so the issues seem to be common:
||JIRA||Change||
|CASSANDRA-15487|iterations from 100000000 to 10000|
|CASSANDRA-9989|perThreadTrees from 10000 to 100|
|CASSANDRA-9932|from 100000000 to perThreadTrees / 10 (in total 1000)|
It seems to me that the perThreadTrees was unintentionally changed in CASSANDRA-15510, so we should revert to the old value. I've also fixed a few places where the perThreadTrees is not used for the right purpose (like for iterations in the testRandomRangeAndBatches), so the whole {{LongBTreeTest}} was scaled incorrectly.
Please, take a look.
I have got successful results here, but still waiting for the final run:
[https://ci-cassandra.apache.org/view/patches/job/Cassandra-devbranch/2301/flowGraphTable/]
was (Author: mmuzaf):
[~mck], [~benedict],
The reason the test suite hangs is that the number of running tests configured by the {{perThreadTrees}} configuration parameter is too large to complete within a reasonable time (the test suite timeout is 90 minutes). There is no such problem for the {{cassandra-3.11}} branch, as {{perThreadTrees}} is set to 100, so the test completes within 15 minutes. See [LongBTreeTest.java#L61|https://github.com/apache/cassandra/blob/cassandra-3.11/test/burn/org/apache/cassandra/utils/LongBTreeTest.java#L61].
The total number of "runs" has been changed several times to fit the timeout, so the issues seem to be common:
||JIRA||Change||
|CASSANDRA-15487|iterations from 100000000 to 10000|
|CASSANDRA-9989|perThreadTrees from 10000 to 100|
|CASSANDRA-9932|from 100000000 to perThreadTrees / 10 (in total 1000)|
It seems to me that the perThreadTrees was unintentionally changed in CASSANDRA-15510, so we should revert to the old value. I've also fixed a few places where the perThreadTrees is not used for the right purpose (like for iterations in the testRandomRangeAndBatches), so the whole {{LongBTreeTest}} is scaled incorrectly.
Please, take a look.
I have got successful results here, but still waiting for the final run:
https://ci-cassandra.apache.org/view/patches/job/Cassandra-devbranch/2301/flowGraphTable/
> LongBTreeTest times out after btree improvements from CASSANDRA-15510
> ---------------------------------------------------------------------
>
> Key: CASSANDRA-18131
> URL: https://issues.apache.org/jira/browse/CASSANDRA-18131
> Project: Cassandra
> Issue Type: Bug
> Components: Local/Memtable
> Reporter: Michael Semb Wever
> Assignee: Maxim Muzafarov
> Priority: Normal
> Fix For: 4.0.x, 4.1.x, 4.x
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Happening in both ci-cassandra.a.o and circleci.
> LongBTreeTest is timing out on 4.0, 4.1, trunk branches.
> Started back in mid April (https://github.com/apache/cassandra/commit/018c8e0d5e and https://github.com/apache/cassandra/commit/596daeb7f08).
> Nightlies shows when the failures started, evident by the 'jdk=jdk_1.8_latest,label=cassandra,split=7/' subfolder missing in the following…
> - https://nightlies.apache.org/cassandra/trunk/Cassandra-trunk-test-burn/1254/Cassandra-trunk-test-burn/
> - https://nightlies.apache.org/cassandra/cassandra-4.0/Cassandra-4.0-test-burn/343/
--
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