Originally posted by: tkilchenmann

Hello,

I need to expand a PowerHA Shared Volume Group with additional hdisks. Doing that I get an error stating, that max. PP per PV is will be exceeded.

On a standalone system I would run chvg -t <factor> <vg> for workaround. There is no such option in the smitty hacmp cspoc menues.

Now I am not sure if I can run chvg outside of cspoc without risking unplanned outage of service/powerha or/and PowerHA configuration errors. So I need your experience advice with that topic.

Thanks in advance for your help.

Timo

Originally posted by: j.gann

Ciao Timo,

if the shared vg is varied on in enhanced concurrent mode you'll need to vary it off and on in standard mode before using chvg. this means either stop the RG or temporarily remove the vg from the running RG.

cheers

Joachim

Originally posted by: tkilchenmann

Hallo Joachim,

Many thanks for you answer. Very appriciated.

Timo

Originally posted by: bodily

A nice side affect of ecvgs (when online in active/passive on the nodes as appropriated) is many lvm, though not jfs info specifically, attributes are automatically updated to the other node(s). I honestly don't remember if the factor size is one or not.

Having said that if you make a change, and/or didn't use C-SPOC (or in this case you can't use C-SPOC), in a cluster aware fashion, then you use the C-SPOC option to update shared volume group (menu name may vary a bit based on version). 

OR you can simply run the C-SPOC CLI of cli_updatevg (located in /usr/es/sbin/cluster/cspoc) with the proper syntax and that will pull in all changes. This is my preferred method. This became available in 5.5, and I think back ported to 5.4.1, but also in 6.1 and 7.1 as well.

Though I completely understand Joachims comments, that has been the norm in the past, but really you shouldn't ever have to do that as it causes at least a node/partial cluster outage to do so, which of course is frowned upon.

Hope this helps.

Shawn

Originally posted by: garyBennett

I'm running PowerHA 6.1 and have also been told in an IBM PMR that we have to stop the cluster and varyon the VGs in normal mode to expand LUNs (chvg -g).  In fact, the "PowerHA for AIX Cookbook", last updated in late 2010, suggests this as well on pg 328.

Could you share the basic procedure for a chvg on a running ECVG in a PowerHA cluster?  Would be greatly appreciated.

TIA.

gary

Originally posted by: POWERHAguy

This has been doable for quite a while now. I'm not sure which AIX levels started allowing it. But the man page for chvg is usually pretty good about updating information on such limitations.  Also during a very recent update to the Cookbook you reference, I made sure that section was  updated to reflect.

Too many times, though little fault of their own, support line continues to use old procedures and unaware of these updates. Mainly because they get the answer from the PMR database and its luck of the draw if the one they find has the newer way of doing it.

I have a demo on doing it, though it shows latest levels of PowerHA.

https://www.youtube.com/watch?v=iUB7rUG1nkw

Shawn Bodily
Senior Consultant - Clear Technologies

PowerHAguy on Youtube and Twitter