VMware Cloud Community
Johnny_Utah
Contributor
Contributor

Statistics and an unmanageable database

Hello Vmware people,

Here's some background info:

Virtual Center 2.5 using MS SQL 2005 SP2. 5 ESX hosts and roughly 200 virtual machines.

Last night I ran out of disk space on the database volume, and I noticed that the DB itself had reached 100+ GB. I dropped the level of statistics logging down from 4 to 2, and ran this query to blow away old stats:

http://www.dennisvanleur.nl/kb/index.php?option=com_content&view=article&id=124:how-can-i-purge-old-...

These deleted about 25 GB worth of data. Not as much as I was hoping, but a good start.

This morning I woke up and saw that overnight, the database had grown back to its original size, eating up the 25 GB of free space.

Can anybody shed some light as to what the hell is going on here?

Thanks,

Johnny

Reply
0 Kudos
4 Replies
cainics
Enthusiast
Enthusiast

Check on your existing Virtual Center Statistics setting and let me know what it states for 200 VM's. It should definitely not say 100 Gb or so and should be way less.

Secondly, what is happening in your environment. Is there a lot of log information being written to the VC logs. Check the logging options and change it to "Warnings and Errors" if it is set to a more detailed logging.

Most organizations/sys admins only need to document and act on errors/warnings. Generally other information is not needed (depending on the nature of business).

Let's use this as a starting point.

Thanks

“Your intelligence is measured by those around you; if you spend your days with idiots you seal your own fate.”

Reply
0 Kudos
Johnny_Utah
Contributor
Contributor

Thanks for the reply.

With the statistics level at 2, 5 physical hosts, and 200 virtual machines, it show 0.96 GB.

Logging was set to information, I changed it to warnings and errors.

Do you think the logging set to info could have really filled up 100+ GB?

Thanks,

JEsse

Reply
0 Kudos
T3Steve
Enthusiast
Enthusiast

I had a similar issue. With over 100 hosts and 2000 vms in my VCDB, the DB size is under 6GB. I once had the stats level set too high and had 1.6 BILLION records in the VPX_HIST_STAT1 table in about a week. This filled up the DB drive and caused VC to crash. I ran the database cleanup script and it woudl purge only 14 million records in 8 hours. Eventually I just manually purged that table and haven't looked back.

VCP3|VCP4|VSP|VTSP
Reply
0 Kudos
Johnny_Utah
Contributor
Contributor

Thanks - Yeah, I just noticed that same table is 25+ GB. The SQL query from Vmware doesn't appear to be very effective at all.

Did you just run a delete on the table name to remove the statistical data?

Reply
0 Kudos