RedHat kernel problems with LVM snapshots

[ScalixSupport]

We are aware of a number of issues with RedHat kernels relating to kernel hangs or panics when using LVM snapshots.

On FC4, using kernel-2.6.8-1.541 or above, attempting to create a snapshot generates a kernel error. This is raised as RedHat bug 132057
As of 1st December 2005 there is no indication of any imminent fix.

There is a corresponding RHEL4 bug for the same issue 164959
As of 1st December 2005, a fix will be available in RHEL 4 Update 3.

On RHEL3, after installing Update 6, attempts to mount a snapshot volume cause a hang. This is raised as RedHat bug 171983.
As of 1st December 2005, there is no indication of any imminent fix.

We advise the following:

1. For a new installation, use RHEL3 Update 5. This is the most tried and tested RedHat version that Scalix customers are using.
2. Customers who have installed RHEL3 Update 6 do not need to back anything out provided that they use the kernel that was available in RHEL 3 Update 5.
3. In the case of FC4, the only alternative is to stop the Scalix server and take an offline backup.

Best regards

Scalix Support Team

[pete]

It appears (to me) that the RHEL 3 bug mounting a snapshot only occurs if the underlying filesystem is EXT3. I have worked around this by explcitly mounting the snapshot as type EXT2 (mount -o ro -r ext2 ....). While this does cause some kernel messages to be syslogged, the snapshot is usable.

Works for me - YMMV

P

[btisdall]

Hmmmm... snapshots *seem* to be working fine here on FC4 using kernel 2.6.14-1.1637_FC4 and mounting as ext3. Haven't actually tried using the snapshotted data in anger yet but no kernel errors.

One issue I did have is that I first started using snapshots, lvcreate -s failed with a usage error until I realised that the dm_snapshot module wasn't loaded at bootup, I had to modprobe it.

--
Ben Tisdall

[jch]

The snapshot problem only arises for some people. I had no luck at all reproducing it in the lab. Conversely, some customers have had no luck at all taking a snapshot. The key feature seems to be the amout of notionally free memory compared to the size of the volume you're taking the snapshot of.

The bottom line is that if it works for you, great, if it doesn't you'll need to hang on for the RHEL4 update 3 kernel. If the timing of preevious patches is anything to go by (and I don't work for Red Hat, so this is only a guess); RHEL4 U3 should go into beta some time in the next two or three weeks and be released properly about six weeks later. Thanksgiving and Christmas are likely to much up the guesses though so I wouldn't be all that surprised if the beta didn't appear until January and since there aren't any published schedules (at least not published where I can see them!) I can't even be sure about that.

jch

[NoelProf]

The RHEL4 update 3 kernels are out. This update definitely seems to have fixed our problem with the system hanging during the LVM snapshot process. We have a very small amount of free extents to allocated extents. I think that contributed to us running into this bug with the Redhat LVM system.

[jch]

Thanks very much for confirming that the fix works!

jch

[mhanisch]

Hello,

just as a reference for others:

We're experiencing similar problems, but using RHEL 4 Update 4.
These occur only sporadically, but often enough to give us trouble.

A new ticket has been opened in Red Hat's bugzilla - have a look at
https://bugzilla.redhat.com/bugzilla/sh ... ?id=215237 if you're interested.

So far, we have no idea about how to fix this and have not heard back from RH, but we also haven't been able to find other people reporting similar problems.
If anyone has experienced the same troubles I'd be glad to hear from you.

[jch]

There are some good suggestions in the bug now as to how to go about tracking this one down. You've raised a bug in bugzilla but have you also raised a support ticket with Red Hat? (Anyone can put bug reports in bugzilla, but you have to pay to be able to raise support tickets.) I await the results of the suggestions in the bug with interest though.

The original snapshot problems were caused by lack of memory and the fix addressed that issue by basically pre-allocating the memory for the snapshot and that makes the comment in the bug about memory particularly pertinent.

jch

[ng]

hello,

does anybody know if this has been fixed or at least confirmed by red hat or someone else?

nikolaus

[dkelly]

Looking at the bug, they are still waiting for the original submitter to include some extra information. If you are experiencing the same thing, I would suggest that you fill in some of the blanks for RedHat.

Cheers

Dave

[kjakkanen]

This is an old sticky, but I can understand its position if someone's had problems with RHEL 4.4 which is not THAT old.

We did run LVM snapshots successfully with RHEL 4.4 and currently working with at least Update 5.

-Kimmo