HPUX v2 cluster upgrade to v3

[njia_1]

Hi everyone

I am planning an upgrade of few (rx2660 and rx8640)clusters running SG 11.18, HPUX v2 to HPUX v3.

It is an MC environment and we are only allowed to take one node down a time.

my plan is
halt pkg on primary node
start pkg on standby
halt SG service on primary
install HPUX v2 on top of existing OS
finger crossed
then the new node join the existing cluster and pkg moved back to primary node

I believe HPUXv3 and HPUXv2 can work together within one cluster is that correct ?

Has anyone came cross similar situation but got a better idea ?

anyone upgraded HPUXv2 to v3 in a MC environment ?

thanks

[Raj D.]

Make sure SG is same version after v3 upgrade , and before joining the new upgraded node to the cluster. if SG in same version it will work with HPUXv3 and v2.

Check out this document:

http://docs.hp.com/en/B3935-90096/ch01s04.html


Hth,
Raj.

[Suraj K Sankari]

Hi,
If SG version is same then no issue what is the OS version you have, but it is suggested that you need to upgread your OS ASAP for both the node.

Suraj

[njia_1]

Hi there

thanks for your reply, just checked the cluster

8+1 rx8640 cluster running SG 11.17 HPUX v2 so I think we need to upgraded the SG sfotware as 11.17 does not support HPUX 11iv3 (11.17.01 does)

I guess we will need to upgarde SG first then HPUX v3.

7+1 rx2660 cluster running SG 11.18 HPUXv2

but I got confused by the HPUX v3 minimum firmware matrix, it says rx2660 server needs to have at least 01.05 for HPUX v3 and 02.31 for HPUX v2, how come HPUX v2 needs a more recent FM revision than HPUXv3 ?

here is the sysrev command output from one of the rx2660 we have and I guess what we have is not supported on hpux v2 ?

SYSREV

Current firmware revisions

MP FW : F.02.23
BMC FW : 05.24
EFI FW : ROM A 07.14, ROM B 07.14
System FW : ROM A 04.11, ROM B 04.11, Boot ROM A
PDH FW : 50.07
UCIO FW : 03.0b
PRS FW : 00.08 UpSeqRev: 02, DownSeqRev: 01

[njia_1]

ahh I must be looking at the wrong firmware version, just noticed it the System Firmware, so what we have is 04.11, that should be fine.

I think I got enough info to start, will build test cluster and perform the upgrade.

thanks

[Raj D.]

njia,

> 8+1 rx8640 cluster running SG 11.17 HPUX v2 so I think we need to upgraded the SG sfotware as 11.17 does not support HPUX 11iv3 (11.17.01 does)

- Yes you have to do SG rolling upgrade.
On the servers having low SG version , you have to upgrade the SG first, then to proceed with OS upgrade.
- One more , before SG upgrade you have to check the firmware compatibility for 11iv3 upgrade. You can run the msv2v3check utility and provide the output to HP for H/W F/W support check.

> 7+1 rx2660 cluster running SG 11.18 HPUXv2

-In that case you must choose to have 11iv3 with SG 11.18 version. ( sept 2007 11iv3 OE comes with SG 11.18 ).


If you want to consistant SG on all the boxes, you can upgrade SG 11.17 and SG 11.17.01 to 11.18,
Then plan for OS upgrade.


Hth,
Raj.

[Raj D.]

njia,

Also check this documents:
- And 11i v2 to 11i v3 Mass Storage Check Utility (msv2v3check )



1. HP-UX Server Support Matrix (PDF)
[ http://h20338.www2.hp.com/hpux11i/downloads/public_hp-ux_systems_support.pdf ]



2. 11iv3 : Mass storage Support Matrix :
http://www.hp.com/products1/serverconnectivity/support_matrices.html#hpux11iv3


3. msv2v3check utility:
https://h20392.www2.hp.com/portal/swdepot/displayProductInfo.do?productNumber=MSV2V3Check



cheers,
Raj.

[Benoy Daniel]

Both server need to be same OS in order to run a two node cluster.

[Wim Rombauts · Accepted solution]

Hello njia,

I have completed the very same procedure as you have just a few months ago, and it all went very well.
We are running Oracle RAC, on this cluster and a set of other softwares, so it was a heavily used cluster with a lot of packages.

What did I do :

1) Upgrade ServiceGuard to 11.19

Why : Current HP-UX 11i v3 comes with ServiceGuard 11.19
This is a rolling upgrade. Bring all packages to the other node, install 11.19 (including recent patches !) and start again. You can run a mixed 11.18 and 11.19 cluster as long as you don't go reconfiguring your cluster while they have a mixed version.
Also, take care there are no packages that start or fail back automatically to the server you just upgraded. It is VERY imoportant that the cluster is fully formed before any package is moved, specifically when the last server is upgraded.

At the end, run a cmapplyconf to verify your configuration file is still valid (possible outdaterd configuration paramaters).

2) Upgrade the first OS

This is also a rolling upgrade once your SG is alredy 11.19
I have done a cold install, not an upgrade.
Most software and data is on external storage, using LVM, and I backed-up all relevant local data that I need to retain. (ssh configuration and keys, firewall configuration, kerberos configuration, APA and LAN-card configuration, ...)
As for ServiceGuard, just copy /etc/cmcluster from the partner server to the server you just upgraded and run cmapplyconf.

Start your system, start your cluster and move all applications tio the upgraded system.

3) make an Ignite backup of your first system, and use this to cold install your second system in the cluster.

Just change IP-addresses, hostnames, ssh-keys, .... and you are back in business.

We were, anyway.

4) I had to upgrade 3 2-node clusters, but all cluster nodes had a 95% similar configuration. I only cold installed the first system, untill I was happy with the configuration. All other servers were installed by Ignite, and just needes minor modifications to run.

[njia_1]

Hi Wim Rombauts

Looks like in one stage your two node cluster are running different version of HPUX, is that correct ? did you have any problems with that ? as you can see from the earlier replies some said for a 2 node cluster both nodes must have same OS.

one more question regarding your cold rebuild of the node, did you just copied /etc/cmcluster (including the binary cluster configuration) to the HPUX v3 node and then the HPUX v3 node just joined the cluster ?

thanks