cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Joel Knighton (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-12436) Under some races commit log may incorrectly think it has unflushed data
Date Thu, 11 Aug 2016 19:25:20 GMT


Joel Knighton commented on CASSANDRA-12436:

The problem here is with the 3.0 branch. Unfortunately, it is concealed by a few unfortunate

The [3.0 unit tests|] didn't actually
run because unit test compilation failed. There is a bug in some of the test tooling such
that it doesn't preserve return codes, so a build that fails on compilation will run no tests
and appear as passing (since it has no failed tests). I'm doing my best to get this fixed
soon and may do it myself.

On the 3.0 dtests, some ran successfully because of the order of compilation (C* -> C*
unit tests -> C* tooling). Since the bug was in C* unit tests, stress didn't get built
so the class was missing.

The problem is in the unit test added. CDC introduced the segmentManager field. It was called
allocator in previous versions, including 3.0. In addition, the {{executeFormattedQuery}}
method was added in [CASSANDRA-12282], which hasn't been committed to 3.0.

> Under some races commit log may incorrectly think it has unflushed data
> -----------------------------------------------------------------------
>                 Key: CASSANDRA-12436
>                 URL:
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Branimir Lambov
>            Assignee: Branimir Lambov
>             Fix For: 3.0.9, 3.9
> This can mainfest itself as a "Failed to force-recycle all segments; at least one segment
is still in use with dirty CFs." message after CASSANDRA-11828.

This message was sent by Atlassian JIRA

View raw message