problem with kernel patches 051 ? [message #26138] |
Wed, 16 January 2008 07:54 |
arghbis
Messages: 35 Registered: November 2007
|
Member |
|
|
Hello everybody,
I come here to report a strange situation, and i don't know if it's due to ovz kernel or to something else, i'm just trying to understand and solve this problem.
I've begun a thread yesterday about strange memory management ( http://forum.openvz.org/index.php?t=tree&goto=26088& #msg_26088).
Today, i'd like to report another problem, so here is the story :
1 - a machine have been attacked and compromised
2 - we have isolated it and made a complete "from scratch" reinstall
3 - however, for openvz, we have used the same template and same VE configs
4 - we also use the same softwares, using the same releases and the same content
Now, we have huge iowaits, a rm -rf on a temp directory takes about 2 minutes were it used to take less than 20sec, and we feel that the server is less responsive than before.
The only differences i see is that it runs 2.6.18-ovz-028stab051.1-enterprise on reiserfs partitions while, before reinstall, it ran 2.6.18-ovz-028stab048.1-enterprise on ext3 partitions.
And now, it comes to my mind that i have tested another server and drop it due to huge iowaits that induced huge load average (about 1300 points of load, yes, you read correctly) due to iowaits. For this last one, i thought it was an hardware problem, since i tested several kernels (2.6.18 / 2.6.22, either vanilla patched with ovz latest patch, or ovz debian-systs kernel). But know that i see the new problems, i think that there also may be a kernel problem.
So, does anybody have already observed such problems, or think to something that can explain this?
thanks for any opinion or clue to solve this
|
|
|
Re: problem with kernel patches 051 ? [message #26232 is a reply to message #26138] |
Thu, 17 January 2008 22:19 |
khorenko
Messages: 533 Registered: January 2006 Location: Moscow, Russia
|
Senior Member |
|
|
Hi,
may be, just to prove if the new kernel is guilty, you can just reboot the node back to the old kernel and look after the node load?
--
Konstantin.
If your problem is solved - please, report it!
It's even more important than reporting the problem itself...
|
|
|