3

We have VCenter Server 4.1 set up to use Oracle database (11r2 with the latest patches).

Everything works fine except the vcenter server starts very slowly, at the startup time one of the oracle processes eats near 100% of 1 CPU, it happens about 5-10 minutes every startup then almost no any CPU load.

There couldn't be a network issue - both VCenter and Oracle run as VMs on the same ESXi box.

We have 2GB RAM allocated for Oracle but EM shows it's more than enough - it never eats more than 1GB.

When the same VCenter used it's bundled MS SQL Standalone, everything started up fine. The database is small - a couple of the test servers in one cluster and about 40 virtual machines running on them.

VCenter OS: Win 2003 x64, Oracle OS: OEL5

Is everybody facing the same problem with VCenter+Oracle?

2
  • How is your SGA/PGA allocated in the oracle instance? What are the startup parameters in the init.ora (or spfile)?
    – REW
    Oct 28, 2010 at 4:41
  • memory_target=1G processes = 150 open_cursors=300
    – disserman
    Oct 28, 2010 at 18:36

4 Answers 4

1

It's really not what you want to hear but we're a big Oracle client but had to give up with our ~30 VC boxes when we moved them to Oracle due to similar odd problems (we had a lot of DB disconnections too). We've not had any problems since moving back to MSSQL, it's not that I'm a big fan either, it seems to be more stable.

Hope you get a better answer than this but I just wanted you to be aware of our problems.

1
  • 1
    thank you, at this moment the problem is only with a slow startup. I know the MS SQL is more suitable, especially for the small configuration but as I wrote it's rather a test VSphere cluster than a production and we need to have a stable solution on Oracle or make a verdict for our customers to avoid it at all :)
    – disserman
    Oct 13, 2010 at 18:11
1

11gR2 is not supported by VMware at this time.

1

http://www.vmware.com/files/pdf/techpaper/vsp_41_perf_VC_Best_Practices.pdf

In vCenter Server 4.1, add a Unique Index in the VPXV_DEVICE_COUNTER table for better performance. create unique index VPXI_DEVICE_COUNT_IDX on VPXV_DEVICE_COUNTER(entity_id, device_name, stat_id);

0

I would double check the resources tab in both virtual machines and make sure the "unlimited" box is checked for CPU and Memory.

Also, if they are in a resource pool, try taking them out of the resource pool and put them in the root of the ESXi server.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .