OpenVZ Forum: Support » *SOLVED* Raid 10 problem?

Home » General » Support » *SOLVED* Raid 10 problem?

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

*SOLVED* Raid 10 problem? [message #10815]

Sun, 04 March 2007 15:29

mperkel
Messages: 253
Registered: December 2006

Senior Member

Running into a problem and not sure what I'm doing wring. Created a software raid 10 array. Everything seems to be normal except that if you take the array down and run e2fsck on it there are always errors, mostly all little stuff and it recovers without losing any data.

I just upgraded to the latest 2.6.18 kernel and that didn't fix it.

The problem will happen most all the time. I can format the partition, put the data on it then shut it down and run 2efsck and there will be some errors to fix.

In the past after running for weeks the raid array will unexpectedly go inti read only mode due to the errors. After running e2fsck it will run normal again.

I've never used raid 10 before (stripes on top of 2 mirrors) so I don't have anything to compare this with. I'm just wondering if I'm doing something wrong. I am using ACL and dir_index attributes.

If I run e2fsck twice in a row it comes up with no errors he second time. However if I mount the /dev/md2 device and then immediately umount it and run e2fsck again I get some errors.

e2fsck 1.39 (29-May-2006)
Pass 1: Checking inodes, blocks, and sizes
Inode 17172774, i_blocks is 8304, should be 19272. Fix? yes

Inode 37126149, i_blocks is 872, should be 2192. Fix? yes

Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
Block bitmap differences: +(39500534--39501904) +(74267806--74267970)
Fix? yes

Free blocks count wrong for group #1205 (1371, counted=0).
Fix? yes

Free blocks count wrong for group #2266 (17206, counted=17041).
Fix? yes

Free blocks count wrong (218321825, counted=218320289).
Fix? yes

/vz: ***** FILE SYSTEM WAS MODIFIED *****

125791 inodes used (0.11%)
451 non-contiguous inodes (0.4%)
# of inodes with ind/dind/tind blocks: 7086/648/0
15831007 blocks used (6.76%)
0 bad blocks
3 large files

107783 regular files
11529 directories
1584 character device files
10 block device files
3 fifos
1 link
4872 symbolic links (4864 fast symbolic links)
1 socket
--------
125783 files
*

Junk Email Filter
http://www.junkemailfilter.com

[Updated on: Mon, 05 March 2007 16:07] by Moderator

Report message to a moderator

Re: Raid 10 provlem? [message #10818 is a reply to message #10815]

Sun, 04 March 2007 17:14

rickb
Messages: 368
Registered: October 2006

Senior Member

Hi. When the fs goes into read only mode, what does "dmesg" say? If there are reports of block device problems, the problem is much different than if there is not.

From my experience, if you are seeing seek errors in dmesg, its a problem with the disks/controller however if there are no block level errors/reports, it is a problem with the md driver.

This may help you narrow it down a lot.

-------------
Common Terms I post with: http://wiki.openvz.org/Category:Definitions

UBC. Learn it, love it, live it: http://wiki.openvz.org/Proc/user_beancounters

Report message to a moderator

Re: Raid 10 provlem? [message #10819 is a reply to message #10815]

Sun, 04 March 2007 17:20

mperkel
Messages: 253
Registered: December 2006

Senior Member

I don't have the messages anymore but I do remember that there were a lot of errors and that was the reason it switched to read only. I'm using software raid and using the sata ports on the motherboard. The errors weren't hardware errors though.

I just don't think it's hardware because I just don't think that merely mounting and then unmounting the array shoule leave these kind of errors.

Junk Email Filter
http://www.junkemailfilter.com

[Updated on: Sun, 04 March 2007 17:23]

Report message to a moderator