InnoDB log buffer unflushed is wrong

[GoogleCodeExporter]

The InnoDB log buffer unflushed in the InnoDB Log graph shows spikes that
reach about twice as large as the log buffer size.  This is impossible,
since the unflushed bytes can't be larger than the log size.

Screenshot is attached.

Original issue reported on code.google.com by baron.schwartz on 25 May 2008 at 6:13

Attachments:

• graph_image_024.png

[GoogleCodeExporter]

Should $status['unflushed_log']
         = max($status['unflushed_log'], $status['innodb_log_buffer_size']);

be min() instead for now then?

Original comment by graphae...@gmail.com on 11 Aug 2008 at 6:21

[GoogleCodeExporter]

I think this must be related to issue 36 and issue 41

Original comment by baron.schwartz on 8 Nov 2008 at 4:44

[GoogleCodeExporter]

the unflushed_log rrd is a COUNTER, it should be a GAUGE!

Original comment by fri...@gmail.com on 13 Nov 2008 at 3:39

[GoogleCodeExporter]

There is sponsorship for the fix in Percona customer issue 7963.

Original comment by baron.schwartz on 16 Dec 2009 at 1:59

[GoogleCodeExporter]

This is fixed in the template definition file in r418, but it's not really fixed
until there are instructions about upgrading and fixing the RRD file on the 
wiki.

Original comment by baron.schwartz on 24 Jan 2010 at 5:35

• Changed state: Started

[GoogleCodeExporter]

Removing unused labels.

Original comment by baron.schwartz on 16 Apr 2010 at 2:14

• Removed labels: Priority-Medium

[GoogleCodeExporter]

This issue was closed by revision r479.

Original comment by baron.schwartz on 21 Apr 2010 at 2:07

• Changed state: Fixed