[prev in list] [next in list] [prev in thread] [next in thread]
List: veritas-ha
Subject: Re: [Veritas-ha] Oracle and cron jobs
From: Asim Zuberi <asim () colltech ! com>
Date: 2000-11-29 15:54:21
[Download RAW message or body]
Here is how I am doing:
#!/usr/bin/ksh
#
#
# The first part of the script determines, if the node within the
# cluster is active host or not.
#
FO_IF="qfe1:1"
ACT_IF=`ifconfig -a | grep qfe1:1 | awk '{print $1}' | sed
s/:1:/:1/g`
if [ "$ACT_IF" == "$FO_IF" ] ; then
continue
else
exit 0
fi
#
# you can key in your stuff from down here...
#
# END
hope this helps...
Asim;
On Tue, 28 Nov 2000 17:38:26 Skip Jones wrote:
> I'd be interested to hear from anyone else, but here's my two cents
> worth.
>
> I've given this much thought over the years really - nothing unique about
> this situation to VCS. One thought I've had but haven't fully explored
> is the idea of a separate job scheduling system that could be put under
> HA control. While using such a third party solution would ease the
> problem of ensuring the job metadata migrates when the application moves
> to a different node (if done properly) you're still left with the
> following conditions that really should be checked for and corrected (and
> I think the only real place you can safely to this is within the scripts
> themselves):
>
> 1. Is the Service Group that is suppose to be running on this node
> actually running?
>
> 2. You need to account for jobs that abort. What if the node fails when
> a job is running? In many cases this isn't a problem, but in some cases
> you might need to have your jobs clean up after themselves when they
> restart.
>
> I think one flaw with your desire to offload your cron jobs to a non HA
> node is that you've now introduced a single point of failure into your
> application landscape. Perhaps in your environment you guys can survive
> without the scheduled job runs for some amount of time?
>
> Anybody else got any ideas?
>
> -Skip
>
> On Tue, 28 November 2000, Chris Smith wrote:
>
> >
> >
> > Hey guys,
> >
> > I'm working on setting up VCS for the first time, and was wondering how
> > people deal with cron jobs on the different servers.
> >
> > Here's our situation - we have a bunch of oracle databases that are
> going
> > to move from server to server. Currently, the oracle user on each
> server
> > has entries in cron to do such various things as kick off backups, do
> > exports and imports, send mail, etc. Now, in this brave new clustered
> > world, we're not going to be able to assume that a particular database
> is
> > on any given machine. So far, we've come up with a few solutions:
> >
> > * Distribute a single cron file to all the servers and have each job
> check
> > to see what databases are running before doing anything.
> >
> > * Have canonical versions of all the crontab files on each server and
> > modify each service group's online script to copy the correct one
> into
> > place (or append to what's already there).
> >
> > * Set up a VCS-controlled filesystem that moves with the databases that
> > contains all the scripts. Then have a central server (not in the
> > cluster) that connects to the IP address associated with each
> database
> > and kicks off the scripts remotely.
> >
> > I personally prefer the last option, the problem being that all the
> > scripts have paths & such hardcoded this would require a major
> re-write.
> > I know that we can't be the first to deal with this, so what have you
> all
> > done?
> >
> >
> > Thanks!
> >
> > Chris
> >
> > _______________________________________________
> > Veritas-ha maillist - Veritas-ha@mailman.eng.auburn.edu
> > http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha
>
>
> _______________________________________________
> Veritas-ha maillist - Veritas-ha@mailman.eng.auburn.edu
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha
>
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic