Portal Home > Knowledgebase > Articles Database > High io usage and server hang
High io usage and server hang
Posted by adamsgr, 05-09-2012, 05:43 AM |
I have 3 dedicated server all with software raid-1 sata, centos or cloudlinux and cpanel.
To all of them the last 10 days or so during cpanel incremental backup server has high io usage and hangs out.
No server is overloaded with many account's (10 accounts -60 accounts and 300 accounts average each server).
To the one with 300accounts ( ex-6 hetzner) i added a 3rd hd for /backup didn't see any change.
I could check for hardware failure but to have the same issue to 3 servers and be hardware failure i would hit jackpot
Any ideas to search for the issue?
Attached Thumbnails
|
Posted by ArturasLIX, 05-09-2012, 06:57 AM |
Is it kernel panic or just irresponsible for a while? What about error logs, output of top or whatever at hung time?
|
Posted by adamsgr, 05-09-2012, 07:51 PM |
No it is not kernel panic just irresponsible.I didn't see any weird thing at error logs.
Attached Thumbnails
|
Posted by ZW-Brooke, 05-09-2012, 08:07 PM |
I asked the techs here and the first thing that came to mind for them was possible upcoming hard drive failure. They recommended that you use the command-line tool for your RAID controller to monitor the status of the drives closely.
They aren't 100% sure that its hardware failure because of the fact the IO load isn't evenly distributed, but they say something as little as a slightly bad SATA cable could cause such IO wait under load as well.
Hopefully this information helps in some way. :-)
|
Posted by adamsgr, 05-09-2012, 08:22 PM |
Thanks a lot that was my first impression was that, i will put one server down for checking but as i said there are 3 dedicated server's with the same problem, i think it is quite odd to be happening.
|
Posted by NineConnect, 05-09-2012, 10:29 PM |
I would expect cPanel backups to cause some I/O issues on a RAID 1 box with normal web server activity on the server as well. You may want to try and use "ionice" for changing the I/O scheduling priority for the problem process/service. As you have mentioned, it's most likely not bad hardware if all servers are affected.
|
Posted by adamsgr, 05-13-2012, 08:23 PM |
I have found the source of the problem. Was the logaholic mysql dump. I disabled it and things are normal again.There is a thread in cpanel forums about it
http://forums.cpanel.net/feature-req...st1156971.html
|
Add to Favourites Print this Article
Also Read