Originally posted by: IBMFORUMS

Hi All

how many volume groups HACMP can handle?
I have a cluster of two nodes when it start with more than 9 VGs, get crashed and reboot the server.

Message was edited by: IBMFORUMS
#AIX-Forum

Originally posted by: alethad

I don't know of any limitation like this in HA. Maybe RG's but not Vg's other than what the OS is limited to. Have you checked the HACMP Info Center docs?

You should have errors somewhere in your error logs. Anything in the syslog or the hacmp.out? Where or when exactly is the system crashing? Does HACMP start at all? No crash code in the LED?
I think you need to give a little more info on this.
#AIX-Forum

Originally posted by: IBMFORUMS

I have this error when I put more than 9 VGs on the cluster:

LABEL: DSI_PROC
IDENTIFIER: 9D035E4D

Date/Time: Thu Feb 21 14:54:48 SAUST 2008
Sequence Number: 3962
Machine Id: 00C869CF4C00
Node Id: localhost
Class: S
Type: PERM
Resource Name: SYSVMM

Description
DATA STORAGE INTERRUPT, PROCESSOR

Probable Causes
SOFTWARE PROGRAM

Failure Causes
SOFTWARE PROGRAM

Recommended Actions
IF PROBLEM PERSISTS THEN DO THE FOLLOWING
CONTACT APPROPRIATE SERVICE REPRESENTATIVE

Detail Data
DATA STORAGE INTERRUPT STATUS REGISTER
0000 0000 0000 0000
SEGMENT REGISTER, SEGREG
0000 7FFF FFFF D000
DATA STORAGE INTERRUPT ADDRESS REGISTER
0000 0004 024D 287D
EXVAL
0000 0000 0000 000E

<hr />

and then system dump is created.

however, when the RG is 9 VGs and below, the cluster works fine and no errors in the log. I searched the doc and hacmp site for limitation in the number of VGs, no clue about it.
#AIX-Forum

Originally posted by: alethad

DSI's are usually a device issue more commonly device drivers but you really need to get the crash dump analyzed by IBM if you can. They will be able to tell you quicker what the problem really is than anyone here can. We may have the same error but the devices are personal to your machine.

Good luck.

Message was edited by: alethad
#AIX-Forum

Originally posted by: Carlyle

I have just finished configuring a cluster with 21 volume groups, 11 on one node and 10 on the other. No, there is no restriction but it becomes difficult to oversee. Multiple resource groups is better.

When you sync and verify (smit cm_ver_and_sync.select), what do you see? Are the volume groups properly imported on the partner nodes? Are the major device numbers consistent?

Are there and clues in /var/adm/cluster.log?
#AIX-Forum

Originally posted by: IBMFORUMS

I have one RG with 25 VGs and the other with 9 VGs each on one node.

with sync and verify everything OK.

the 9 VGs RG starts normally.
the 25 VGs RG can't start until I reduce it to 9 VGs.

I have dump I will send it to IBM and see what's the feedback.
#AIX-Forum

Originally posted by: Carlyle

And you haven't found any clues in cluster.log or hacmp.out?
#AIX-Forum

Originally posted by: Carlyle

SYSVMM?

How much memory do you have available. How much paging space have you defined?
#AIX-Forum

Originally posted by: IBMFORUMS

two LPARs with 10 CPU and 25 GB memory
paging 16 GB

Message was edited by: IBMFORUMS
#AIX-Forum