Red Hat Bugzilla – Bug 635706
[RFE] Let's pack ABRT logs as well
Last modified: 2010-11-21 20:51:35 EST
Description of problem:
The ABRT log of various crashes might provide quite useful information about a system that is being troubleshooted.
There exists an ABRT sos plugin which creates a sos report on each crash -- this looks a bit odd to me (very expensive resource-wise) and also having a sum of crashes on an misbehaving system might be an useful information
Version-Release number of selected component (if applicable):
sos svn checkout
Steps to Reproduce:
1. run sos
2. try to find an ABRT log
It's not there
It is e.g. in sos_commands/abrtlog/abrt-log
This needs a new plugin
Created attachment 448477 [details]
Something like this.
When placed in /usr/lib/python2.6/site-packages/sos/plugins/ it does the magic. Please consider adding such a plugin also to the upstream sos package.
This package has changed ownership in the Fedora Package Database. Reassigning to the new owner of this component.
Created attachment 460437 [details]
New patch with checkenabled()
Added checkenabled() using the (admittedly unusual for Python code) "return True; return False" pattern, and a copyright notice.
Tomas, are you OK with your name being included like that?
On a more general note, I had a look at the various options available.
We could even go further and gather infos+backtraces, with something along the lines of (to be Pythonized):
for ID in $(sqlite3 /var/spool/abrt/abrt-db 'select UUID from abrt_v4')
abrt-cli -ib $ID
I don't have a strong opinion for or against it, and would be happy to get feedback on that.
Elements if we want to go further with the idea from my comment 4:
1) abrt-cli doesn't offer any clean way to list the UUID unless we want to parse the 'abrt -lf' output
2) No Python for the sqlite3 request as far as I'm concerned, as Red Hat doesn't ship Python support for sqlite3 in RHEL6;
3) However it seems safe to rely on /usr/bin/sqlite3 in this plugin.
Ran under RHEL6:
# rpm -q --requires abrt|grep sql;rpm -qf /usr/lib64/libsqlite3.so.0; rpm -qf /usr/bin/sqlite3
Thanks again Tomas.
Including your name, it will be removed upon request.
Cloning this bug for a potentially more complete report.
Requested feature implemented through:
thanks for stepping up. The patch provided in the comment #0 was meant more as an illustration and I actually thought that having the full backtraces in the sosreport would be the next step. I didn't want to touch the database directly though -- in parallel I started patching abrt-cli but never got the will to finish it. I will try harder.
TBH: It sounds silly to have my name in a copyright of a file where the biggest part has been written by somebody else. :) However if it's meant as "you're responsible for this", then OK, I will try to help.
Perfect Tomas, thanks a lot for the "burden of authorship".
I agree we should get rid of the sqlite3 approach as soon as possible.
However it's not bugging me that much. IMHO we should file a feature request against abrt-cli, and from our "consumer" PoV, the feature would need to be introduced significantly before anyone ships a different database schema..