Spurious innodb_io_capacity-sized flush list flushes on a loaded server

Bug #1238039 reported by Laurynas Biveinis on 2013-10-10
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
MySQL Server
Unknown
Unknown
Percona Server moved to https://jira.percona.com/projects/PS
Fix Released
Medium
Laurynas Biveinis
5.1
Invalid
Undecided
Unassigned
5.5
Invalid
Undecided
Unassigned
5.6
Fix Released
Medium
Laurynas Biveinis

Bug Description

The furious flushing, implemented by skipping page cleaner thread sleeps, in combination with short-duration LRU and flush list flushes, means that server activity might be not updated by the next page cleaner thread iteration.

This has an effect of spuriously issuing innodb_io_capacity-sized flush list flush request instead of flushing as loaded server is supposed to, which in turn causes performance instabilities.

The fix is to maintain server activity accounting based on 1 second granularity. That is, an active server may only be considered inactive if no server activity bump in the last 1 second, as opposed to last cleaner thread iteration. An inactive server may be considered active immediately after the server activity bump.

Related branches

tags: added: xtradb
tags: added: innodb upstream
removed: xtradb

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PS-1437

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.