[MDEV-4830] Isolation / Rejoin instability using MariaDb and Galera Created: 2013-08-01  Updated: 2017-09-25  Resolved: 2017-09-25

Status: Closed
Project: MariaDB Server
Component/s: Galera
Affects Version/s: 5.5.29-galera
Fix Version/s: 5.5.33a-galera

Type: Bug Priority: Major
Reporter: Stefano Simonelli (Inactive) Assignee: Seppo Jaakola
Resolution: Fixed Votes: 0
Labels: galera
Environment:
  • The tests was done on 3 Virtual Machines with MariaDB-Galera on a Openstack/Devstack enviroment hosted in a SkySQL server.
  • CentOS release 6.3 (Final)
  • Configuration 1
    MariaDB-Galera-server.x86_64 - 5.5.29-1
    galera.x86_64 - 23.2.4-1.rhel6
  • Configuration 2
    MariaDB-Galera-server.x86_64 - 5.5.31-1
    galera.x86_64 - 23.2.6-1.rhel6

Attachments: Zip Archive IsolationRejoin_TestsReport_V03.zip    

 Description   

The tests consisted on successive stop/start and isolate/rejoin commands executed on 1 node, repeated 5-10 times.

The stop/stat test went fine, but we have observed instability of the functionalities testing the isolate/rejoin, sometimes it works and sometimes it doesn’t.

 Comments   
Comment by Seppo Jaakola [ 2013-08-05 ]

Setting wsrep_provider=none hangs in mariaDB build.

Problem seems to be that closing slave applier thread does not broadcast COND_thread_count. Strange thing is that MySQL version works well, although the relevant code path seems to be identical.

MySQL version bug is tracked here: https://bugs.launchpad.net/codership-mysql/+bug/1208493

Comment by Seppo Jaakola [ 2013-08-05 ]

Fix for MariaDB branch pushed in revision: http://bazaar.launchpad.net/~maria-captains/maria/maria-5.5-galera/revision/3410

Generated at Thu Feb 08 06:59:31 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.