Bug 470216 - EXT4-fs error (device xvda5): ext4_get_group_desc: block_group >= groups_count - block_group = 1, groups_count = 1
Summary: EXT4-fs error (device xvda5): ext4_get_group_desc: block_group >= groups_coun...
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Red Hat Enterprise Linux 5
Classification: Red Hat
Component: kernel-xen
Version: 5.3
Hardware: x86_64
OS: Linux
medium
medium
Target Milestone: rc
: ---
Assignee: Eric Sandeen
QA Contact: Martin Jenner
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2008-11-06 10:10 UTC by Alexander Todorov
Modified: 2009-01-10 16:25 UTC (History)
3 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2009-01-10 15:56:06 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)
/etc/fstab (532 bytes, text/plain)
2008-11-06 10:10 UTC, Alexander Todorov
no flags Details
/var/log/dmesg (8.51 KB, text/plain)
2008-11-06 10:10 UTC, Alexander Todorov
no flags Details
/var/log/messages (16.96 KB, text/plain)
2008-11-06 10:12 UTC, Alexander Todorov
no flags Details

Description Alexander Todorov 2008-11-06 10:10:22 UTC
Created attachment 322687 [details]
/etc/fstab

Description of problem:
The message in summary is observed in dmesg while system boots

Version-Release number of selected component (if applicable):
kernel-xen-2.6.18-121.el5

How reproducible:
Not sure

Steps to Reproduce:
1. Install PV xen guest on x86_64 with RHEL 5.3 snap #1 using the attached disk layout.
2. Boot the system
3.
  
Actual results:
error message when system boots

Expected results:
no error message

Additional info:
xvda5 is /home
after doing touch /home/test as root the file is created but a message is printed on the console:
JBD: barrier-based sync failed on xvda5 - disabling barriers

Comment 1 Alexander Todorov 2008-11-06 10:10:57 UTC
Created attachment 322688 [details]
/var/log/dmesg

Comment 2 Alexander Todorov 2008-11-06 10:12:14 UTC
Created attachment 322689 [details]
/var/log/messages

Comment 3 Chris Lalancette 2008-11-06 10:24:17 UTC
Well, I'm pretty sure our PV disk implementation doesn't support barriers (we had a long discussion about it a few months back).  Are barriers required for EXT-4 to function?  Also, is EXT-4 going to be supported in 5.3, or is still going to be tech preview?

Chris Lalancette

Comment 4 Alexander Todorov 2008-11-06 11:05:29 UTC
ext4 will be tech preview in 5.3. Other questions I can't answer.

Comment 5 Chris Lalancette 2008-11-06 11:12:29 UTC
OK, then it shouldn't be a blocker for 5.3.  We'll have to track down what else is going on here; if it turns out we do need barriers, that can be done, it will just have to be 5.4.

Chris Lalancette

Comment 6 Rik van Riel 2009-01-10 01:13:00 UTC
ext4 cannot require barriers, because LVM does not support barriers either.

As for what the ext4 message means, maybe Eric Sandeen knows.

Comment 7 Eric Sandeen 2009-01-10 05:32:10 UTC
ext4 does not require barriers.

JBD: barrier-based sync failed on xvda5 - disabling barriers

is just telling you that they can't be used.

It's also not the cause of this problem; barriers won't matter 'til you crash and have to replay the logs.

FWIW it's a terribly confusing message.  It should be more like:

ext4_get_group_desc: block_group >= groups_count;  block_group = 1, groups_count = 1

Anyway, fixed upstream by:

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=565a9617b2151e21b22700e97a8b04e70e103153

and we'll likely be rebasin ext4 for 5.4 ...  Is this a small, single-bg filesystem?

Comment 8 Rik van Riel 2009-01-10 15:56:06 UTC
OK, so all the warning does is tell you the filesystem has only one block group.  I'll drop this one from the kernel-xen bug list.

Comment 9 Eric Sandeen 2009-01-10 16:25:35 UTC
Rik, well, it does mark the fs with errors, which is somewhat problematic.  But it's not a terribly big deal other than the "mounting filesystem with errors, fsck recommended" message. A single-bg filesystem is pretty rare too, I think.

-Eric


Note You need to log in before you can comment on or make changes to this bug.