Sanity Check Request

Posted by bbrennan on 13-Jul-2016 14:21

OE11.3.3 on Windows 2008 Server - helping a customer with a third party db.

Customer has a database and bi file on a single disk.  The bi is a single fixed extent of 512 meg and a variable extent.  Currently the variable extent is 1.4 gig.  The disk has about 105 gig free out of 275 gig.  The database machine is part of a VMWare configuration.

At 11 this morning, the database went down.  The DB.lg file said that there was no more space to write to the BI file.

Just verifying that the total space for a BI can exceed 2 gig when you combine extents.

If so, I assume that either a process consumed all of the now-free 105 gig or that VMWare lied and some other machine cause the disk space to be used.  This is mystery #1.

Then half an hour later, the database was back running again.  The customer uses OE Explorer.  As far as we can tell, no one restarted the db.  The machine did not reboot and has an up-time of several days.  I have little experience with Explorer so wanted to check that there is no auto-restart feature.  Mystery #2.

Sort of spooky on both accounts.  Thoughts welcomed.


All Replies

Posted by Paul Koufalis on 13-Jul-2016 14:33

Yes 11.3.3 does support total BI size > 2 GB.

Enterprise or Workgroup DB? The latter does not support large files so if any extent hit 2 GB then the DB will go down.

What it smells like is thin provisioning of disk space by VMWare: Progress asked the OS for space to grow the BI, the OS asked VMWare and VMWare said "Urkk, gack, ugghhh...".

AFAIK there is no auto-restart feature in OEE. Look for some script in Scheduled Task that checks and runs dbman -start.

Posted by bbrennan on 13-Jul-2016 14:42

Good to know I have been paying attention.   They run Enterprise.  I was not sure but I did not think there was some 2 gig limit on the total BI extents.  Thanks for that clarity.  I agree with your thinking the lack of disk space was a VMWare issue at the time of the failure.

The restart is still a total mystery. I am confident there are no scripts or scheduled tasks.  I have left this to their internal folks to sort out.  I am just a guest helper for this one.  I'll report back what they find since it is a curiosity.


Posted by James Palmer on 13-Jul-2016 15:20

I assume large files is enabled on the DB?

I concur in regard to no auto-start feature. The DB log file will tell you when the DB started spooling up. Is it a random time or a specific time? If it's random then the plot thickens, if it's a fairly specific time then it definitely points to some sort of task.

Is it possible the start script was actually run on a different machine? dbman allows you to start databases remotely. We do it all the time on our dev environment as we don't want any Tom/***/Harry poking around on the DB server, but they need to be able to start/stop databases.

This thread is closed