Bug 1369123

Summary: logs: sos fails to collect the logs from journalctl
Product: Red Hat OpenStack Reporter: Gonéri Le Bouder <goneri>
Component: sosAssignee: Lee Yarwood <lyarwood>
Status: CLOSED DUPLICATE QA Contact: BaseOS QE - Apps <qe-baseos-apps>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 8.0 (Liberty)CC: agk, bmr, gavin, goneri, lyarwood, mburns, plambri, pmoravec, qe-baseos-apps, sbradley
Target Milestone: ---Keywords: Triaged
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: 1369122 Environment:
Last Closed: 2017-01-19 09:49:39 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Bug Depends On: 1183244, 1369122    
Bug Blocks:    

Description Gonéri Le Bouder 2016-08-22 14:03:59 UTC 
+++ This bug was initially created as a clone of Bug #1369122 +++

Description of problem:


Version-Release number of selected component (if applicable):
sos-3.2-36.el7ost.2.noarch (this is the version from the RHOS8 puddle).


How reproducible:
sosreport -o logs -v (output attached)

100% reproductible

I can extract the log manually:
[root@director tmp]# time bash -c 'journalctl --all --this-boot --no-pager > all'
real    1m26.669s
user    1m9.138s
sys     0m13.246s
[root@director tmp]# ls -lh all
-rw-r--r--. 1 root root 1.4G Aug 22 09:00 all
Comment 2 Gonéri Le Bouder 2016-08-22 14:16:54 UTC 
reducing the size of the logs is enough to avoid the problem (journalctl --vacuum-size=200M)
Comment 3 Mike Burns 2016-10-12 13:38:53 UTC 
Is this an sos problem (in which case, it's RHEL, not OSP). Or is this some default setting that we need to change?
Comment 4 Gonéri Le Bouder 2016-10-12 13:54:06 UTC 
This is an sos problem that impact RHOSP. Should I close it?
Comment 5 Bryn M. Reeves 2016-10-12 14:05:34 UTC 
It's both, really: sos should handle commands that generate large amounts of output on stdout better (open RFE, quite a big job), and it would help if OSP had a default log rotation regimen that would mean sos users are not impacted by this in default configurations & with the default sos options.

We added a new Plugin API method in 3.3, add_journal(), that gives plugin classes better control over what they are collecting - so far the openvswitch, openstack_sahara, docker, logs, and kubernetes plugins are using it. This allows plugins to override the default timeout, start/end time and other aspects of journal collection (although I don't think anyone's using those features, yet).