On Monday 20 October 2008 09:35, van Sleeuwen, Berry wrote:
>Indeed, if the application has the correct init scripts and the SIGNAL
>is trapped to a "shutdown -h now" then a SIGNAL would correctly shutdown
>the application and the guest. But only if the SIGNAL has been given
>enough time to shutdown before CP will force the user.
>
>That would trigger my question, how to determine what the correct time
>would be? We started with 300 seconds, and two years ago we increased
>the time to 600 secs. But we have discovered that even 5 minutes could
>well be too short to shutdown the database.
I ran into this very problem on my product. Originally, we had written it to
do a "shutdown -h now", wait until the network connection terminated (sshd
was stopped), then wait a few minutes more before logging off the guest. But
customers pointed out that this method did not ensure that everything was
shut down before the logoff. If your filesystems haven't been sync'd before
the logoff, you've got a possibility of corruption.
So we now have Provisioning Expert monitor the console output from that guest,
wait until it sees the message saying that the processor has halted, and then
log off the guest. This is basically what Berry suggested: automating what
we would do as admins. It's really the only way because you can't predict
the timing of anything in a virtual environment because you don't control how
much CPU you'll get.
So the problem may not be that the Oracle shutdown sequence hasn't completed
by the time CP logs your guest off. That might finish just fine, but unless
your filesystems are sync'd and unmounted before the logoff, you haven't
really saved the final state of the system.
- MacK.
-----
Edmund R. MacKenty
Software Architect
Rocket Software, Inc.
Newton, MA USA
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
