Red Hat Bugzilla – Bug 29160
Package does not automatically restart crond after upgrade
Last modified: 2007-04-18 12:31:42 EDT
From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows 98; ZDNetSL) The vixie-cron-3.0.1-61.i386.rpm released last week as an upgrade to the RH package does not automatically restart the cron daemon after upgrading. Reproducible: Always Steps to Reproduce: 1.Install package 2. 3.
I can verify this I put this upgrade on 3 RH70 boxes and the results were the same as the original reporter. one machine was reboot for other reasons so it started up crond just fine. But the other two I had to issue a restart. I suspect some post install glitch.
This nailed me, too. RH 7.0 system, and I foolishly went several days without realizing that crond was dead. This has had some ramifications now, which are mostly my fault since I *definitely* should have detected that crond was dead. Nevertheless, please fix the RPM! Thanks.
It stopped it but didn't start it? That's odd. I've just tried on two machines here; I can't duplicate this behavior. Each time the cron daemon does get restarted.
Tough call to make on the erratic behavior, but my crond process died like a foot fungus in a pool of Tenactin.
I've tried on various machines to verify this and I can't. What do the log messags from cron look like around the time the package is upgraded?
No messages, really. Just "Feb 20 10:49:25 mysys crond: crond shutdown succeeded." Now, this is immediately preceeded by the message for me su-ing to root to run "rpm -Fvh vixie-cron..." That's it. I have some software (Mailman, which I noticed that Bugzilla is run with) that runs a script from cron every minute. The entries cease at 10:49 and never start again until today when I figured out what was up and restarted crond.
What does 'rpm -V vixie-cron' say?
The following are from logcheck: Feb 22 10:01:05 system1 sudo:...COMMAND=/bin/rpm -Fvh vixie-cron-3.0.1- 61.i386.rpm *** Upgrade initiated Feb 22 10:01:07 system1 crond: crond shutdown succeeded ***Begin upgrade Feb 22 12:21:12 system1 crond: crond shutdown failed *** Restarting crond, notice stop failed, it was not running Feb 22 12:21:13 system1 crond: crond startup succeeded *** Restarting crond These entries are from the cron log just before the upgrade and when I manually restarted crond: Feb 22 10:01:01 system1 CROND[32720]: (root) CMD (run-parts /etc/cron.hourly) Feb 22 12:21:12 system1 crond[762]: (CRON) STARTUP (fork ok)
Well, "rpm -V vixie-cron" produces no output...literally.
OK, it appears in both cases that it's shutting crond down but not starting it. This is really odd. What version of initscripts do you have installed?
I've got initscripts-5.49-1 (5.49).
Same info as dan@tri-8.com. initscripts-5.49-1 and no info from -V vixie-cron
Same as previous poster. initscripts-5.49-1 and no output from "rpm -V vixie-cron".
I've got another big problem, which I probably will need to submit a new bug for. With all this talk about crond, check this out: now that I've got this new vixie-cron-3.0.1-61, my crons are all running incorrectly. I'm getting *duplicate* runnings, but not every time. For instance, one cron entry runs at "0 23 * * *", which is 11:00 each night. Well, according to my cron logs, the command was issued at 10:59:59 and then *AGAIN* at 11:00:00 (one second later). This was a backup software command, so the backup got all messed up because two instances were fired up at almost the same time. I looked in the cron logs, and my "minutely" mailman cron is doing the same thing: it runs at xx:xx:59 and then at xx:xx+1:00 (one second later), but not every time. Some times it just runs at xx:xx:00. I'm about 4 minutes away from getting the old version of vixie-cron off of the CD and putting it back. Can anyone tell me what's up!?
In response to crontab errant behavior, I have experienced none and logs are clean. And new jobs that I have submitted to crontab execute as excpected.
After upgrading vixie-cron-3.0.1-61, all /var/log/cron.* are empty. It looks like some of the messages were sent to /var/log/messages.
Are you still seeing this with recent RHL releases?
Well, I don't think so. Can't even remember if I've specifically upgraded cron separately on newer releases (I do apply all of the released errata). Afraid I can't help much on that one.
I can't reproduce this with current rawhide at least. I suspect that it is probably ok in recent releases too. Please re-open if there is still a problem.