rodean@hpfcdc.HP.COM (Bruce Rodean) (04/10/90)
In article <4022@plains.UUCP> lodin@plains.UUCP (Joe Schmo) writes: > After starting a cron on one of my discless nodes, I started receiving > this message. It comes from a time that ckpacct is executed on the > discless node, one minute after the server. Am I doing > something wrong? I think I have all the proper CDF's to do this. > > > ln: cklock and cklock1 are identical > > > ************************************************* > Cron: The previous message is the standard output > and standard error of one of your cron commands. > > > If you can help, post or email to: This is my understanding of the problem and others may have a better explanation. The problem is due to system clock synchronization differences between the cnode and the server. The cnode goes into a long wait for the cron event to occur. When it wakes up, it schedules the event, in this case, ckpacct. While it has been sleeping, the cnode clock has been readjusted with the server. Cron thinks that the time for the cron event has not yet happened so it schedules it again. Thus you see the ckpacct error message since the first ckpacct is still running. The workaround is to schedule a more frequent cron command, say /bin/true, every 15-20 minutes or so on the cnode. Then cron won't wake up and find time radically different from what it thinks it is. I think the normal setup for ckpacct is every hour. Bruce Rodean rodean%hpfclg@hplabs #include <std.disclaimer> # No warranty of any kind implied or otherwise