What are recommended Disk settings for a DB server? [message #50709] |
Wed, 16 October 2013 16:01 |
mylesw
Messages: 5 Registered: September 2011 Location: Phoenix AZ
|
Junior Member |
|
|
We have a Proxmox 3.1 HN box with OpenVZ container support. We are sharing a number of CTs against a single CT that is a database server. We run Firebird 1.5 Super Server on CentOS 5 (64 bit) for the database.
In previous OpenVZ versions, we had the odd problem but since migrating to this later PM version, we are seeing an increasing trend of 'Disk seek error' or 'Bad Block' errors, coming from the database server. We've backed up & restored working databases to new containers, only to find the problems re-appear.
This morning it dawned on me that the problem is likely to be related to how disk I/O is being allocated between contains on the HN and if I can set priority to be given to this CT, I could likely stop these errors. I think that the problem is that the database server goes to write or access disk resources, but that they are not as available to it as it would hope and rather than waiting, its timing out. So I was thinking that by prioritizing the CT to have more access to disk, it should remove this issue.
We can change CPU units and have done that, which appears to help. But has anyone here put a database server with some decent user load in a OpenVZ container and tweaked the disk settings to get performance for it over other containers on the HN?
If so, care to share the settings you used and what specific settings you changed to get the performance/reliability boost here?
Thanks in advance for any information.
Myles
|
|
|
|
|