This is a strange one.
First let me say that I believe this issue to be caused by poor application/database design and I am trying to convince the developer that this is the case with some 'hard' evidence but there is one Proxmox related thing I am seeing that is bothering me.
Our Proxmox box runs the latest 1.5 (problem existed in 1.4 also). It is not a very fast io box as it has poor disk performance as per pveperf.
We have a few light vps running on there, one which has a rogue mysql/php application running on it. The largest filled table in that application has a max of 20000 records in it (logging table) and the main data tables have less than 5000 records in them.
We have tried all the my.cnf tweaks and have enabled slow query logging which does show quite a bit of slow queries. However most of the performance variable ratios show reasonable values. Also when the mysql is restarted, this application takes forever to initialize.
At seemingly random instances also at mysql restart, the mysqld service on the vps spikes to at times between 100-300% cpu and causes the vps and host loadavgs to raise (top output). This mysqld process is run by user mysql.
If I switch to the host itself and run top, that same process shows up as being run by postfix. Thats the thing I don't understand and I want to be able to properly explain it to my developer.
After a while, the vps mysqld process will calm down and everything will work normally until the next cycle. I am running other vps with different mysql/php applications that are not affected on the same host. Also nothing seems to be showing up in the logs on the vps about this at all.
Any help would be appreciated. I will make a new vps from scratch, install the application and copy all data there to see if the problem goes away or not.
First let me say that I believe this issue to be caused by poor application/database design and I am trying to convince the developer that this is the case with some 'hard' evidence but there is one Proxmox related thing I am seeing that is bothering me.
Our Proxmox box runs the latest 1.5 (problem existed in 1.4 also). It is not a very fast io box as it has poor disk performance as per pveperf.
We have a few light vps running on there, one which has a rogue mysql/php application running on it. The largest filled table in that application has a max of 20000 records in it (logging table) and the main data tables have less than 5000 records in them.
We have tried all the my.cnf tweaks and have enabled slow query logging which does show quite a bit of slow queries. However most of the performance variable ratios show reasonable values. Also when the mysql is restarted, this application takes forever to initialize.
At seemingly random instances also at mysql restart, the mysqld service on the vps spikes to at times between 100-300% cpu and causes the vps and host loadavgs to raise (top output). This mysqld process is run by user mysql.
If I switch to the host itself and run top, that same process shows up as being run by postfix. Thats the thing I don't understand and I want to be able to properly explain it to my developer.
After a while, the vps mysqld process will calm down and everything will work normally until the next cycle. I am running other vps with different mysql/php applications that are not affected on the same host. Also nothing seems to be showing up in the logs on the vps about this at all.
Any help would be appreciated. I will make a new vps from scratch, install the application and copy all data there to see if the problem goes away or not.