Bug 845259 - [lvmetad] Daemon serving corrupted message to client
[lvmetad] Daemon serving corrupted message to client
Product: Red Hat Enterprise Linux 6
Classification: Red Hat
Component: lvm2 (Show other bugs)
Unspecified Unspecified
unspecified Severity unspecified
: rc
: ---
Assigned To: Petr Rockai
Cluster QE
Depends On:
Blocks: 845269
  Show dependency treegraph
Reported: 2012-08-02 09:34 EDT by Marian Csontos
Modified: 2013-01-02 07:34 EST (History)
9 users (show)

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
: 845269 (view as bug list)
Last Closed: 2013-01-02 07:34:17 EST
Type: Bug
Regression: ---
Mount Type: ---
Documentation: ---
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Attachments (Terms of Use)
the message served by lvmetad (2.79 KB, text/plain)
2012-08-02 09:37 EDT, Marian Csontos
no flags Details

  None (edit)
Description Marian Csontos 2012-08-02 09:34:06 EDT
Description of problem:
I experienced a segfault in pvs, very likely caused by corrupted message served by lvmetad.

There were about 19 instances of pvs with different options running at the same time.

Likely a race condition in lvmetad. The issue is not reproducible, though it affected two clients running at the same time both of them got response with large part of metadata replaced by two characters.

Version-Release number of selected component (if applicable):

How reproducible:
low reproducibility

Steps to Reproduce:
1. test was running multiple pvs with various combinations of -o and -O
Actual results:
incorrect response from lvmetad served to clients

Expected results:
correct metadata should be served

Additional info:
I let running 19 concurrent cycles running pvs for all night and could not reproduce the issue.
Comment 1 Marian Csontos 2012-08-02 09:37:17 EDT
Created attachment 601952 [details]
the message served by lvmetad

I reviewed the routine receiving data from socket, and could not find a problem there, thus I believe the data vere served by lvmetad.
Comment 3 Marian Csontos 2012-08-02 10:30:48 EDT
Remaining pvs running at the same time were either finished correctly or were killed by test.

One more pvs command was issued afterwards and returned correct result.
Comment 4 Marian Csontos 2013-01-02 07:34:17 EST
Issue is not reproducible - I have not seen it since reported.

Also since Bug 845269 was fixed the severity of the bug is low.

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