Comments on: Building a Sun Cluster using Solaris 10, on VMware Server

By: Arth

Arth — Wed, 17 Sep 2008 15:59:01 +0000

Thanks.
I have been suffer the pm_tick delay problem too for several years.
Today I proved your solution and it seems work fine.
Eternal gratitude.

By: Luciano

Luciano — Wed, 20 Aug 2008 16:30:50 +0000

Hi,

I’m using the VMWare Server 2.0 RC1, and my node Solaris 10 release 11/06 with pacthes of DVD EIS at 24/Jun/2008.

When the node 1 start, it display messages,
Aug 20 13:29:20 zeus genunix: [ID 313806 kern.notice] NOTICE: pm_tick delay of 5149 ms exceeds 2147 ms
Aug 20 13:29:21 zeus last message repeated 1 time
Aug 20 13:29:21 zeus genunix: [ID 313806 kern.notice] NOTICE: pm_tick delay of 5150 ms exceeds 2147 ms
Aug 20 13:29:21 zeus genunix: [ID 313806 kern.notice] NOTICE: pm_tick delay of 5149 ms exceeds 2147 ms

And very, very slow.

Please, are You know resolve this question?

By: Ken Chua

Ken Chua — Tue, 19 Aug 2008 06:05:11 +0000

I realized that the problem was because there no vmnet assigned to the ethernet interface. It is working now, thanks for your help and advice.

By: geoff

geoff — Thu, 14 Aug 2008 07:43:15 +0000

You should have at least 3 NICs. If you have configured 3 NICs in VMWare for your guest, but can’t see them in the host, they’re probably not plumbed.

The easy way to plumb them all is to type in (in Solaris):

ifconfig -a plumb

However, from memory, cluster would prefer them not to be plumbed, it will configure the interfaces in the installation of the cluster software.

You will need at least 2 NICs setup for the Virtual Network between hosts as interconnects (which is one of the first things mentioned you need to configure in the documentation above).

You should additionally configure a NIC for the public network.

The 2 NICs on each host that connect to 2 NICs on the other host should not all be on the same network, they will need to be NIC 1 -> NIC 1, NIC 2 -> NIC 2 between the hosts. They are used for heartbeats, global file systems and other magic.

By: Ken Chua

Ken Chua — Wed, 13 Aug 2008 04:59:24 +0000

How do you configure the additional interface for the VM?

I installed the solaris guest VM, but can only “see” one interface e1000g0.

Should I select “Host-only” or “Specific virtual network” for the VM Network Connection settings?

Appreciate your advice. Thanks.

By: geoff

geoff — Tue, 17 Jun 2008 23:59:22 +0000

Mohammad - If you follow the instructions, I hope it will work for you The AMD Processor will need to be of a certain revision for it to work with 64 bit guests.

If it is a relatively new processor, you should be fine.

Let me know if you have any troubles.

By: Mohammad Ali

Mohammad Ali — Tue, 17 Jun 2008 18:20:01 +0000

I got a laptop with windows vista AMD 64 bit dual core processor. Now If I use VMware to install two solaris 10 guest node for Sun Cluster will it work?

Please give me an advice.

By: joe

joe — Wed, 28 May 2008 17:29:00 +0000

Great posting…. But I am having a problem when I try to configure an nfs resource. here is what I get, any ideas?:

***Error***
nfs1b - Failed to analyze the device special file associated with file system mountpoint /nfs1: No such file or directory.

(C189917) VALIDATE on resource nfs-stor, resource group nfs-rg, exited with non-zero exit status.
(C720144) Validation of resource nfs-stor in resource group nfs-rg on node nfs1b failed.

By: ajay

ajay — Tue, 29 Apr 2008 11:52:36 +0000

I found this stuff very good. I am going to do it.

By: geoff

geoff — Thu, 17 Apr 2008 02:13:43 +0000

Hi Jeff,

Thanks for your kind feedback. To help with your problem, lets start here:

Ensure Crash Dumps are Enabled following instructions to enable at:
http://docs.sun.com/app/docs/doc/801-7039/6i1cgngff?a=view

When the boxes panic, it will write out 2 files into /var/crash/[hostname].

Get yourself a copy of Solaris Crashdump Analysis Tool (SUNWscat) http://wwws.sun.com/software/download/products/3fce7df0.html

Run /opt/SUNWscat/bin/scat /var/crash/[hostname]/vmcore.[whatever dump number]

Type in analyze when you get a scat prompt like:
SolarisCAT(/var/tmp/vmcore.0)> analyze

and send me the output of that to start, we may wish to do some more work using SUNWscat later.

Also provide me a copy of your vmx config files for each node.

And last but not least, a /var/adm/messages from both hosts might give us some info too.

The other thing is I haven’t tested this config with Sun Cluster 3.2 Maybe we’ve run into some other grief I have not seen as yet. There is certainly some interesting behaviour with cluster in VMWare.

Email these to geoff at unixsysadmin.net and I’ll see what I can find.

Alternatively, I can provide some ftp space so you can upload the crashdump(s) to me and I can look at them for you.