Red Hat Bugzilla – Bug 782359
Condor HFS quota example returns "Unknown config:" from QMF
Last modified: 2014-11-17 21:24:16 EST
Description of problem: Use configuration example from Grid User Guide: Hierarchical Fair Share (HFS) ----- GROUP_NAMES = group_physics, group_chemistry, group_physics.lab1, group_physics.lab2, group_physics.lab3, group_physics.lab3.team1, group_physics.lab3.team2, group_physics.lab3.team3, group_chemistry.lab1, group_chemistry.lab2 GROUP_QUOTA_DYNAMIC_group_physics = .4 GROUP_QUOTA_DYNAMIC_group_chemistry = .4 GROUP_QUOTA_DYNAMIC_group_chemistry.lab1 = .4 GROUP_QUOTA_DYNAMIC_group_chemistry.lab2 = .6 GROUP_QUOTA_DYNAMIC_group_physics.lab1= .2 GROUP_QUOTA_DYNAMIC_group_physics.lab2= .2 GROUP_QUOTA_DYNAMIC_group_physics.lab3= .6 GROUP_QUOTA_DYNAMIC_group_physics.lab3.team1 = .2 GROUP_QUOTA_DYNAMIC_group_physics.lab3.team2 = .2 GROUP_QUOTA_DYNAMIC_group_physics.lab3.team3 = .4 ----- Restart condor. Go to cumin Grid::Quotas Look into cumin's web.log. 4019 2012-01-17 10:24:08,293 ERROR CallSync: call returned with status of Unknown config: GROUP_QUOTA_group_chemistry 4019 2012-01-17 10:24:08,440 ERROR CallSync: call returned with status of Unknown config: GROUP_QUOTA_group_physics Version-Release number of selected component (if applicable): (RHN) cumin-0.1.4916-1.el6.noarch devel cumin-0.1.5184-1.el5.noarch How reproducible: 100% Steps to Reproduce: 1. Setup HFS quota according GUG 2. Restart condor 3. Go to cumin Grid::Quotas 4. Look into cumin's web.log. Actual results: Cumin produces errors Expected results: Cumin runs without error Additional info: Was part of Bug 760567
This errors aren't noticeable from cumin's web. Web is working as expected - like there were no errors.
From what I can tell, this error is coming through from the QMF method call. That status actually is "Uknown confing: blah". The question is, why does this happen and is it okay/expected? Retargeting to condor-qmf. Note, it is easy to see this from cumin. I haven't tried to reproduce from qpid-tool or another script...
The QMF method impl is literally looking for a config param called "GROUP_QUOTA_group_chemistry" but "GROUP_QUOTA_DYNAMIC_group_chemistry" is what has been configured. Back to cumin.
Looks like cumin lacks knowledge of which groups are static and which groups are dynamic, so it tries all groups in both cases. The error messages are unfortunate but expected and handled correctly, it seems. Without changes on the QMF side Cumin cannot know the members of each group ahead of time. Maybe we need an RFE for this...
Fixed in revision 5223. Changed logging level from ERROR to DEBUG in CallSync when methods return non-zero status. This removes the ERROR conditions, and under INFO level logging no messages will be seen. Applies generally to all methods invoked through CallSync.
Technical note added. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. New Contents: Cause Certain QMF method calls return errors under normal Cumin operation. These errors are not a problem and are expected and handled. Cumin writes these errors in the log files. Consequence Users may be concerned that something is wrong with Cumin/QMF operation when actually there is not a problem. Change Change the logging level of these types of errors to DEBUG so that they may be seen in a development context but will not be seen in a normal user deployment. Result Misleading error messages are removed from the Cumin logs.
Notes on testing: The default log level for a Cumin install is INFO. In a default installation and using the "Steps to Reproduce" from the description, there should be no log messages from "Unknown config": 1. Setup HFS quota according GUG 2. Restart condor 3. Go to cumin Grid::Quotas 4. Look into cumin's web.log, there should be no "Unknown config" log entries (or grep "Unknown config" web.log) Change the log level to debug in /etc/cumin/cumin.conf: [common] log-level: debug 1. Restart Cumin 2. Go to cumin Grid::Quotas 3. Look into cumin's web.log The "Unknown config" messages should be back with a DEBUG level on the log entries.
Reproduced on cumin-0.1.5184-1 Tested on RHEL 5/6 i386/x86_64 cumin-0.1.5658-1 Everything works as expected. --> VERIFIED
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHSA-2013-0564.html