*SOLVED* On a raid1 system, when disk IO is high, The system might become "freeze"... [message #20049] |
Tue, 11 September 2007 15:12  |
xinming
Messages: 1 Registered: September 2007
|
Junior Member |
|
|
Hello,
hmm,
I found this thread
http://forum.openvz.org/index.php?&t=msg&th=3186
is almost the same as what I encountered.
I'll tell what It will be like in these situation. On my gentoo box,
I have raid1 on my root. When I just *REBOOTED*, and I do this.
dd if=/dev/zero of=/test.img bs=1M count=512
During the excution of dd, this will cause the machine almost hang.
We can't either ssh in, nor switch to other terminal.
But after the dd command finished, The machine worked again.
If there is nothing related with disk.
The load would go over 10.00 in top command.
If you need any info, Please tell me.
Note: All what I do is on the host. Not Virtual Environment.
As for now, The disk performance is so poor, That I don't even have patience to install the guest.
This is the result when I do dd. For 5.2 MB/s, It's the *fastest* time I ever meet after using the openvz kenrel. >_< I tested this several times. and the worst performance would be only 2.2M/s
and hdparam tells me that my DMA of the disk is enabled.
/ # dd if=/dev/zero of=/test.img bs=1M count=512
512+0 records in
512+0 records out
536870912 bytes (537 MB) copied, 102.314 s, 5.2 MB/s
Sollution:
Sorry to disturb.
It's solved now, I ever thought It's a openvz kernel modified too much that some part of DMA can't be enabled.
I recompiled kernel with the right modules built-in.
Previously, They're modules, and they are loaded automatically into the kernel.
Thanks for your attention anyway.
[Updated on: Thu, 13 September 2007 07:41] Report message to a moderator
|
|
|
|
|