Re: Recommended settings - Mikael Ronström - com.mysql.lists.cluster

4 messages in com.mysql.lists.clusterRe: Recommended settings

From	Sent On	Attachments
Donovan Francesco	03 Jul 2008 03:29
Mikael Ronström	03 Jul 2008 05:08
Donovan Francesco	03 Jul 2008 05:15
Matthew Montgomery	03 Jul 2008 07:12

Subject:	Re: Recommended settings
From:	Mikael Ronström (mik...@mysql.com)
Date:	07/03/2008 05:08:55 AM
List:	com.mysql.lists.cluster

Hi,

3 jul 2008 kl. 12.30 skrev Donovan Francesco:

Good Day,

I am having endless issues with REDO log for example Got temporary error 410 'REDO log files overloaded, consult online manual (decrease TimeBetweenLocalCheckpoints, and|or incre' from NDBCLUSTER

Can anyone recommend the correct settings i need to have, my config.ini looks like this :

[tcp default] SendBufferMemory=10M ReceiveBufferMemory=10M

[NDBD DEFAULT]

NoOfReplicas=2 DataMemory=10288M IndexMemory=3072M

FragmentLogFileSize=16M NoOfFragmentLogFiles=8 UndoIndexBuffer=10M RedoBuffer=256M MaxNoOfConcurrentOperations=100000 #NoOfDiskPagesToDiskAfterRestartTUP=150 #NoOfDiskPagesToDiskAfterRestartACC=83

#Checkpointing...

Diskcheckpointspeed=10M Diskcheckpointspeedinrestart=100M TimeBetweenGlobalCheckpoints=1000

If you're database is 50% full, then it will contain about 6.5GByte of data. This will take 650 seconds to checkpoint. Thus you need a REDO log which can handle about 3000 seconds (including some safety margins). Your current setup has 8 * 64 MByte of REDO log so can handle about 170 kByte of REDO log per second at 50% full and only 85 kByte/sec when close to full.

However it's not uncommon to generate 3-4 MByte of REDO log per second if you have a high write load. Thus you would need to change NoOfFragmentLogFiles to about 200. If you know you have lower load then you can set it lower. You should be able to check how fast REDO log is used with some cluster log event. Don't know which and it isn't enabled by default.

Rgrds Mikael

#the default value for TimeBetweenLocalCheckpoints is very good

TimeBetweenLocalCheckpoints=3

#table related things

MaxNoOfTables=4096 MaxNoOfAttributes=24756 MaxNoOfOrderedIndexes=2048 MaxNoOfUniqueHashIndexes=512

[MYSQLD DEFAULT]

[NDB_MGMD DEFAULT]

My Setup is as follows : 2 x 16GB Ram storage nodes 2 x 4GB Ram application nodes 2 x management nodes.

Please HELP!!!