[PLUG] Possible Causes of Cron Inconsistencies -- FAILURE REPORT

[Rich Shepard]

On Tue, 21 Apr 2009, Larry Brigman wrote:

> resending because of mailing error.

   Is this a pun, Larry? :-)

>> Or is it possible that this is getting caught in a mail filter somewhere?

   I doubt this very much. I used to see reports mailed from root to me
rejected by postfix because the report had a UCE rejection rule in it.
Haven't seen that in several years.

>> There should be a log every time it is run even if there is no mail. The
>> only problem would be detecting the problem if the log file is over
>> written each time the script runs.

   What I did not mention in this morning's continuation of the thread is
that logwatch included it's less useful maillog report.

   Yes, there should be a report generated by 1pflogsumm each time it runs.
There is always content in /var/log/maillog and this rotates each day at
04:40am. The report I included in this morning's message was generated at
00:02am today, so it reflects running 1pflogsumm at that time from root's
crontab. There was not another report run at 04:40am which is when
/etc/cron.daily/1pflogsumm is supposed to be run.

   The total lack of consistency and predictability makes it exceptionally
difficult to try to track down the cause.

   Week before last, neither invocaton of 1pflogsumm ran for days; then both
appeared Wed-Sat. Last Sunday, neither. Monday: only the 00:02 run. Today:
neither. But, the details show it ran successfully as the exit code was 0.

Thanks,

Rich