Bug 192066 - lvcreate -s fails often
Summary: lvcreate -s fails often
Status: RESOLVED DUPLICATE of bug 178321
Alias: None
Product: SUSE Linux 10.1
Classification: openSUSE
Component: Kernel (show other bugs)
Version: Final
Hardware: i686 SuSE Linux 10.1
: P5 - None : Normal (vote)
Target Milestone: ---
Assignee: Jan Blunck
QA Contact: E-mail List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-07-13 07:19 UTC by Peter Münster
Modified: 2007-01-05 16:51 UTC (History)
1 user (show)

See Also:
Found By: Customer
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Peter Münster 2006-07-13 07:19:08 UTC 
With normal SuSE-kernel lvcreate -s often hangs:
killing does not help, even not "kill -9", only reboot

With official kernel-2.6.17.3 lvcreate -s often fails
with the following message:
lvcreate -L 700M -s -n snap-usr -p r /dev/system/usr
  LV system/snap-usr in use: not deactivating
  Couldn't deactivate new snapshot.

Often means: about 50% of times.
The reason is perhaps some race condition in the udev-system.
Comment 1 Hannes Reinecke 2006-07-13 08:02:10 UTC 
Ah. yes. This is the official kernel.
Of course it'll fail.

We have some patches for the device-mapper which introduces 'online' and 'offline' events for device-mapper devices. Problem is that you'll need a whole bunch of device-mapper ioctls to have a useable device.
Unfortunately the device event is sent after the very first ioctl, at which time the device isn't useable at all; you wouldn't even know which _type_ the device-mapper device will have.

Of course, this issue and the patch has been greeted with silence on the device-mapper mailing list. Redhat doesn't believe in device hotplugging :-(

Please try with the standard 10.1 kernel and re-open bug if the problem persists.
Comment 2 Peter Münster 2006-07-13 08:32:32 UTC 
Since with the standard 10.1 kernel lvcreate -s hangs,
which is worse than normal failure, I'll continue to use
kernel-2.6.17.3
Cheers, Peter
Comment 3 Chris L Mason 2006-07-13 14:33:21 UTC 
This is already fixed in the sles kernels, so the next maintenance update for 10.1 will have the fix.  The short explanation is that dm from mainline was full of races for device add/removal.
Comment 4 Greg Kroah-Hartman 2006-07-13 18:03:23 UTC 
Ok, marking as fixed.
Comment 5 Peter Münster 2006-08-21 09:42:59 UTC 
Hello,
now I've installed the kernel-smp-2.6.16.21-0.13:
lvcreate does no more hang, but it fails in the same
manner as the 2.6.17.3 kernel.
Cheers, Peter
Comment 8 Jan Blunck 2007-01-05 16:51:36 UTC 
There were multiple issues with snapshots in the 10.1 release. The kernel side of the hang should be fixed in the SLES10_GA_BRANCH.

The user-space problem (that you describe in comment #0) was fixed in bug 178321 which was a bug in lvm2. Please update to lvm2 version >= 2.02.13 from 10.2 or openSUSE Factory.

*** This bug has been marked as a duplicate of bug 178321 ***