Red Hat Bugzilla – Bug 1142113
As a user, I would like my Pulp logs to go to logical locations
Last modified: 2018-09-19 11:22:40 EDT
Description of problem: /var/log/pulp is full of empty log files for pulp and the workers. All logs go to /var/log/messages or the journal (on rhel 7). We should use /var/log/pulp instead. Version-Release number of selected component (if applicable): 6.0.4 GA
Since this issue was entered in Red Hat Bugzilla, the release flag has been set to ? to ensure that it is properly evaluated for this release.
I don't think pulp upstream should use /var/log/pulp/. We are currently using the standard log facilities given to us by the OS, and we have an explanation for our logging approach here: https://pulp-user-guide.readthedocs.org/en/latest/troubleshooting.html#logging As for the empty log files in /var/log/pulp, we are required by celery to have them. If there is an error during startup of a celery process, before it switches to the system log facility, it will log that error to its file in /var/log/pulp/. All of that said, it could be appropriate for the downstream product to use a rsyslog rule to redirect log messages to a file. There have been multiple previous discussions about why pulp cannot do that upstream. I suggest either closing this bug, or please provide a more detailed request.
Why exactly are they required by celery? Certainly that must be fixable? My preference would be those files go away on rhel 7 and on el6, ship the rsyslog rule to use /var/log/pulp. I don't think this should be closed at all, are those two things really impossible upstream? I've seen lots of Satellite users confused by the presence of these empty log files.
As is, Celery will not start without being given some place to log, and it does not support syslog by default. Believe it or not, it is important that we continue to have these log files as this is where startup errors will be logged. If Celery is unable to start Pulp, error messages about that will go to those files. This is very similar to Apache's error_log. If Apache cannot start Pulp, it will log to /var/log/httpd/{ssl_,}error_log. Once Celery starts Pulp, Pulp will configure logging (i.e., redirect all logging to syslog). I have two ideas to make this clearer to users: 1) We could rename the log files to include the word "startup". For example, reserved_resource_worker-0-startup.log or similar. 2) We may be able to inject one log statement to those log files just before configuring logging that says something like "initial startup successful, redirecting logs to syslog." How does that sound?
Yea, thanks, I think that would work. 1 and 2 together would resolve the issue.
The upstream bug is fixed against pulp 2.6.0, so it will be in sat 6.1.
The Pulp upstream bug status is at CLOSED - CURRENTRELEASE. Updating the external tracker on this bug.
Based on the comments above, moving this to POST. This can be verified against any Satellite 6.1.+ instance.
Verified in Satellite 6.2 Beta Snap 6 Compose 1.
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. https://access.redhat.com/errata/RHBA-2016:1501