Spam in error log if server run without permission to set nice 0

[Laurynas Biveinis]

If server is started in environment where setting nice to 0 is not allowed (e.g. non-root with mysqld niced to priority lower to 0), the error log will have

2014-08-04 15:15:06 28999 [Warning] InnoDB: Setting thread 29686 nice to 0 failed, current nice 10, errno 13
2014-08-04 15:15:06 28999 [Warning] InnoDB: Setting thread 29685 nice to 0 failed, current nice 10, errno 13
2014-08-04 15:15:06 28999 [Note] InnoDB: Percona XtraDB (http://www.percona.com) 5.6.19-67.0 started; log sequence number 5547985478636
2014-08-04 15:15:06 28999 [Warning] InnoDB: Setting thread 29687 nice to 0 failed, current nice 10, errno 13
2014-08-04 15:15:06 28999 [Warning] InnoDB: Setting thread 29688 nice to 0 failed, current nice 10, errno 13

These messages are benign and caused by the thread scheduling priority implementation that attempts to nice every worker thread on startup. If no innodb_sched_priority_cleaner etc option is specified, the default value is 0, which is attempted to nice to here.

A fix would be to add another value to innodb_sched_priority_* options that means "do not attempt to nice" and will be the new default.

[Krunal Bauskar (krunal-bauskar)]

This seems to happen only once at the start of the server and as pointed before message are benign so I would prefer to leave it as is and add a documentation note.

Also, in either case we may need to notify user through a warning/note that we have failed to set the priority as requested.
I think current warning framework look good.

[Laurynas Biveinis (laurynas-biveinis)]

Upstream 5.7 logs the corresponding message at Info level, while our 5.6 does this at Warning level.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PS-3171