I have a small penguin farm running about 100 RH 7.2 Linux images which are
primarily Apache servers under z/VM 4.2.  One of the Linux systems is
support machine I use as a loghost to receive remote syslog data from all of
the other Linux guests. After upgrading the kernel on loghost from the
2.4.9-17 kernel RH 7.2 came with to 2.4.17 (with the timer patch) I noticed
that every morning after the change to the new kernel the time for loghost
was off.  It somehow looses anywhere from 7 - 10 hours every night.  This is
not a time zone shift that is occurring.  The system is off by minutes too.
We are not running a NTP client either.  If I reboot loghost it does come up
with the correct date and time.

My first experiment was to stop the syslog daemon for 1 night to see if the
time was still incorrect in the morning.  With syslog stopped, the system
did not loose any time over night.

I rolled the loghost back to the 2.4.9-17 kernel for now, and the problem is
gone.

I was just reading Martin Schwidefsky's response to a related thread and
wondered if there might be something not being accounted for by the timer
patch which is causing loghost to loose time overnight.  My guess is that
this occurs around 4 AM when lots of cron jobs kick off and the system gets
real busy.

Lonny

_____________________________________
Lonny Sivey
System Support Division
OCLC Online Computer Library Center, Inc.
6565 Frantz Rd, Dublin, OH 43017
(614) 764-6013  FAX (614) 718-7200
mailto:[EMAIL PROTECTED]
_____________________________________

Reply via email to