Sporadic xb_incremental_compressed_* failures in Jenkins

[Alexey Kopytov]

There are sporadic xb_incremental_compressed_* failures in Jenkins builds like the following one:

http://jenkins.percona.com/view/XtraBackup/job/percona-xtrabackup-2.1-param/148/BUILD_TYPE=release,Host=debian6-64,xtrabackuptarget=xtradb55/testReport/junit/%28root%29/t_xb_incremental_compressed_2kb/sh/

2013-01-11 04:41:59: xb_incremental_compressed_2kb.sh: Stopping server with id=1...
130111 4:41:59 [Note] /home/jenkins/workspace/percona-xtrabackup-2.1-param/BUILD_TYPE/release/Host/debian6-64/xtrabackuptarget/xtradb55/test/server/bin//mysqld: Normal shutdown

130111 4:41:59 [Note] Event Scheduler: Purging the queue. 0 events
130111 4:41:59 InnoDB: Starting shutdown...
Warning; Aborted waiting on pid file: '/home/jenkins/workspace/percona-xtrabackup-2.1-param/BUILD_TYPE/release/Host/debian6-64/xtrabackuptarget/xtradb55/test/var/0/mysqld1.pid' after 60 seconds
2013-01-11 04:42:59: testrun.sh: Found a leftover mysqld processes with PID 5695, stopping it

[Alexey Kopytov (akopytov)]

Identical failure in bug514068.sh:

http://jenkins.percona.com/view/XtraBackup/job/percona-xtrabackup-2.0/973/BUILD_TYPE=debug,Host=centos5-64,xtrabackuptarget=innodb50/testReport/junit/%28root%29/t_bug514068/sh/

2013-01-12 20:58:17: bug514068.sh: Stopping server with id=1...
130112 20:58:17 [Note] /home/jenkins/workspace/percona-xtrabackup-2.0/BUILD_TYPE/debug/Host/centos5-64/xtrabackuptarget/innodb50/test/server/bin//mysqld: Normal shutdown

Warning; Aborted waiting on pid file: '/home/jenkins/workspace/percona-xtrabackup-2.0/BUILD_TYPE/debug/Host/centos5-64/xtrabackuptarget/innodb50/test/var/0/mysqld1.pid' after 60 seconds
130112 20:59:17 InnoDB: Starting shutdown...
: Found a leftover mysqld processes with PID 13707, stopping it

[Laurynas Biveinis (laurynas-biveinis)]

Claiming this one due to the issue becoming much more prominent on 5.6.

[Alexey Kopytov (akopytov)]

This might be a server bug which makes the issue more frequent with 5.6. The default timeout for "mysqladmin shutdown" to wait for a pid file to disappear is 3600 seconds, which should be more than enough.

However, for me 5.6 hangs on clean shutdown with "[Note] Giving X client threads a chance to die gracefully" quite frequently. Still using "kill -9" for the test suite is a good idea regardless of this bug.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PXB-614