Please login or register. December 15, 2018, 08:17:50 PM

Author Topic: Adding a vpath to a 2 node Cluster using HACMP  (Read 15328 times)

0 Members and 1 Guest are viewing this topic.

angeek

  • Jr. Member
  • **
  • Posts: 5
  • Karma: +0/-0
Adding a vpath to a 2 node Cluster using HACMP
« on: May 10, 2007, 08:25:44 PM »
Hi,

Can anyone help I need to add a vpath that is already on my system but is not assigned to a volume group yet .

The Aix version I am running is Aix 5.2 ML08 and the version of HACMP is 5.2

What would I have to do to make it synchronise on the standby server when I have it added.

Also is there an easy way I can find out if it is concurrent , cascading or rotating.

I rang the clshowres , and the fallover policy is fallover to the next node in the list. Fallback policy is never fallback . Am assuming this is cascading .

Wondering is there a more accurate way of finding out.

Thanks

Angela


Appreciate some help !!!!
« Last Edit: May 10, 2007, 08:51:23 PM by angeek »

Michael

  • Administrator
  • Hero Member
  • *****
  • Posts: 1101
  • Karma: +0/-0
Re: Adding a vpath to a 2 node Cluster using HACMP
« Reply #1 on: May 11, 2007, 09:04:18 PM »
Definitely going to need to study more about MPIO next week.

However, in terms of HACMP I dont think vpath is a monitored resource by HACMP. If it is, then it would be assignable in the extended resource smit menus. However, probably it is just getting it assigned to the disks in the volume group.

re: cascading, rotating, etc..

Those terms were "retired" starting with HACMP 5.1. From the sounds of it, your resource group policy follow what would have been called a cascading resource group.

Basically, a resource group now has three policies that govern it's behavior: startup, fallover (i call this failover frequently, but fallover is the correct documented term), and fallback.

The default policies are, respecitvely: start on Home node, takeover on next node, fallback to higher priority node (old cascading behavior). Your policy of never fallback means it is "modified cascading". However, if your startup policy is "Start on Any node", then your policies more close resemble "rotating".

Basically, this new arrangement makes it much simplier to specify what is going on, or expected to happen, rather than a "name" followed by a list of exceptions.

Hope this answers most of your immediate questions.
« Last Edit: June 22, 2007, 11:27:37 AM by Michael »

angeek

  • Jr. Member
  • **
  • Posts: 5
  • Karma: +0/-0
Re: Adding a vpath to a 2 node Cluster using HACMP
« Reply #2 on: May 13, 2007, 08:00:15 PM »
Thanks Michael  , that has made it so much clearer.

Micheal I have 2 further questions .

Is there a way of checking the size of a vpath before it is added to the system ?

Also if I need to convert an ordinary volume group to a big volume group  in HACMP can this be done on the fly. The volume group that needs the vpaths added currently has 32 volumes in there which is its limit.

 

Michael

  • Administrator
  • Hero Member
  • *****
  • Posts: 1101
  • Karma: +0/-0
Re: Adding a vpath to a 2 node Cluster using HACMP
« Reply #3 on: May 14, 2007, 09:10:12 PM »
1) dont know about checking the size of a vpath.

2) you can convert to Big on the fly.

regards,
Michael

angeek

  • Jr. Member
  • **
  • Posts: 5
  • Karma: +0/-0
Re: Adding a vpath to a 2 node Cluster using HACMP
« Reply #4 on: June 05, 2007, 08:28:55 PM »
Thanks zoomax for the detailed reply it has proved invaluable.

Can you let me know why a vpath would be locked ?

Raido

  • Registered
  • *
  • Posts: 1
  • Karma: +0/-0
Re: Adding a vpath to a 2 node Cluster using HACMP
« Reply #5 on: June 22, 2007, 08:18:30 AM »
hm... where can I find Zoomax's notes? Those were useful stuffs. :]

angeek: because the reservation in most cases (to avoid multiple access to the same storage)
« Last Edit: June 22, 2007, 08:22:59 AM by Raido »