A quick recap of the problem:
------------------------------------------
We noted that jobs that were scheduled to run at times ending in zero
weren't happening. I asked if this was a known problem and if a patch
was available.
Subsequently, I discovered:
---------------------------------------
The problem was *not* restricted to times ending in zero - it was just
coincidental that the episodes that triggered everything were scheduled
at zero times. A number of jobs were being skipped throughout the day.
Summary of responses:
----------------------------------
1. A couple of folks confirmed that they had seen jobs at times ending
in zero fail to fire off. Both indicated that they simply rescheduled
jobs to run 1 minute later. [Would work, except for later discovery
that other jobs were getting missed, not just times ending in zero. I
suspect that people tend to notice when time zero jobs don't run,
because they are often "sexy" jobs that generate regularly reviewed
output - this was certainly our experience.]
2. Glenn Satchell (glenn@uniq.com.au) had not seen jobs being missed,
but did point me to patch #103690-02, which claims to solve the
following bugs:
Patch-ID# 103690-02
Keywords: cron NIS NFS hang stat initgroups fields crontab
Synopsis: SunOS 5.5.1: cron patch
Date: Dec/03/96
1255421 *cron* cron jobs are not always executed
1178638 *cron* job doesn't run if set to run on 2 fields in fields 3, 4
or 5 of crontab
(from 103690-01)
1230784 *cron* can lose jobs when NIS or NFS servers hang
----------
Doug Stein Voice: 415-968-1101
FirstFloor Software, Inc. Fax: 415-968-1193
444 Castro Street, Suite 200 Email: dstein@firstfloor.com
Mountain View, CA 94041 Web: http://www.firstfloor.com