[SOLVED] Growing redo.log file

[reason2008]

Hello,

I'm running Zimbra Open Source Edition, version 6.0.3. I noticed that there is a redo.log file that has grown to nearly 900MB. I've searched around and haven't found any clear information on how to manage this file. I know the purpose of the file. Is there a script that is supposed to rotate this particular file? What can I do to reduce the file size? Any suggestions will be appreciated.

Thanks!

[mmorse]

The new default is 1GB before rollover/archive (or delete by default in the case of FOSS), you can adjust yours down with zimbraRedoLogRolloverFileSizeKB.

zmprov mcf zimbraRedoLogRolloverFileSizeKB #
or
zmprov ms server.domain.com zimbraRedoLogRolloverFileSizeKB #

The default used to be 100MB but that only allots 10x10MB or 4x25MB etc uploads/messages which is not very many transactions fitting in there so it was increased: Bug 32687 - Change default redolog rollover size

[reason2008]

Thank you very much!

[ewilen]

What does zimbraRedoLogRolloverHardMaxFileSizeKB do?

[reason2008]

Originally Posted by ewilen

What does zimbraRedoLogRolloverHardMaxFileSizeKB do?

It sets the maximum size of the redo.log file. When the file reaches that size, it is copied an archive directory and a new, empty file is created, I believe. With the open source free edition of Zimbra it looks like the file is deleted, or emptied, when it reaches the max size specified.

[ewilen]

I think you just described zimbraRedoLogRolloverFileSizeKB, which is a different but related property. Note the "HardMax" in the name.

[ewilen]

Okay, so I've been running ZCS 6 NE since March. Given the new default for zimbraRedoLogRolloverFileSizeKB, there should only have been 1 redolog/day if the redolog was less than ~1 GB.

So why do I have

/opt/zimbra/backup/sessions/incr-20100516.080011.771/redologs:
total 750M
-rw-r----- 1 zimbra zimbra 569M May 16 01:01 redo-20100515.032453.409-seq1281.log
-rw-r----- 1 zimbra zimbra  15K May 16 01:01 redo-20100515.043038.802-seq1282.log
-rw-r----- 1 zimbra zimbra  27K May 16 01:01 redo-20100515.050253.580-seq1283.log
-rw-r----- 1 zimbra zimbra 180M May 16 01:01 redo-20100516.080011.695-seq1284.log

/opt/zimbra/backup/sessions/incr-20100528.080014.711/redologs:
total 490M
-rw-r----- 1 zimbra zimbra 387M May 28 01:01 redo-20100527.235836.622-seq1297.log
-rw-r----- 1 zimbra zimbra 2.0M May 28 01:01 redo-20100528.000315.295-seq1298.log
-rw-r----- 1 zimbra zimbra 398K May 28 01:01 redo-20100528.000441.929-seq1299.log
-rw-r----- 1 zimbra zimbra 167K May 28 01:01 redo-20100528.000751.385-seq1300.log
-rw-r----- 1 zimbra zimbra 8.5M May 28 01:01 redo-20100528.001106.311-seq1301.log
-rw-r----- 1 zimbra zimbra 623K May 28 01:01 redo-20100528.001516.318-seq1302.log
-rw-r----- 1 zimbra zimbra 2.8M May 28 01:01 redo-20100528.001758.651-seq1303.log
-rw-r----- 1 zimbra zimbra 4.7M May 28 01:01 redo-20100528.002031.038-seq1304.log
-rw-r----- 1 zimbra zimbra 1.8M May 28 01:01 redo-20100528.002424.187-seq1305.log
-rw-r----- 1 zimbra zimbra 5.8M May 28 01:01 redo-20100528.002746.692-seq1306.log
-rw-r----- 1 zimbra zimbra 2.0M May 28 01:01 redo-20100528.004655.640-seq1307.log
-rw-r----- 1 zimbra zimbra 1.2M May 28 01:01 redo-20100528.005339.768-seq1308.log
-rw-r----- 1 zimbra zimbra 592K May 28 01:01 redo-20100528.010622.964-seq1309.log
-rw-r----- 1 zimbra zimbra  72M May 28 01:01 redo-20100528.080014.617-seq1310.log

/opt/zimbra/backup/sessions/incr-20100530.080011.061/redologs:
total 465M
-rw-r----- 1 zimbra zimbra 304M May 30 01:00 redo-20100528.212855.552-seq1311.log
-rw-r----- 1 zimbra zimbra 161M May 30 01:00 redo-20100530.080011.058-seq1312.log

The only guess I have is that maybe a new redolog is created when a restore is performed; I did a lot of restores on 5/28.

Then on 5/31 I did zmprov mcf*zimbraRedoLogRolloverFileSizeKB 102400. At this point, redologs should generally be about 100 MB each, right? So why do I have

/opt/zimbra/backup/sessions/incr-20100602.080014.694/redologs:
total 1.3G
-rw-r----- 1 zimbra zimbra 479M Jun  2 01:01 redo-20100601.182911.843-seq1315.log
-rw-r----- 1 zimbra zimbra  49M Jun  2 01:01 redo-20100601.185325.165-seq1316.log
-rw-r----- 1 zimbra zimbra  21M Jun  2 01:01 redo-20100601.191131.227-seq1317.log
-rw-r----- 1 zimbra zimbra 259K Jun  2 01:01 redo-20100601.191349.061-seq1318.log
-rw-r----- 1 zimbra zimbra  53M Jun  2 01:01 redo-20100601.193633.326-seq1319.log
-rw-r----- 1 zimbra zimbra 6.8M Jun  2 01:01 redo-20100601.194502.842-seq1320.log
-rw-r----- 1 zimbra zimbra 632M Jun  2 01:01 redo-20100602.080014.690-seq1321.log

/opt/zimbra/backup/sessions/incr-20100603.080012.511/redologs:
total 582M
-rw-r----- 1 zimbra zimbra 582M Jun  3 01:01 redo-20100603.080012.507-seq1322.log

Some more restores were done 6/1 and I'm aware of a mass-mailing with a 2 MB attachment around 8am. I suppose that could have overflowed the limit before seq1315 could be rolled over to seq1316. But why was seq1321 allowed to grow to 632M? And why is seq1322 582M?