[prev in list] [next in list] [prev in thread] [next in thread] 

List:       veritas-ha
Subject:    Re: [Veritas-ha] operations and maintenance tasks
From:       "Brad Willer" <brad_willer () symantec ! com>
Date:       2007-02-06 1:32:28
Message-ID: 07BD096D1B9D1F4692858C276EFBBB0A0B5A546D () svlxchcln5 ! enterprise ! veritas ! com
[Download RAW message or body]

A variation that you can do that expands a little bit on what Eric said
is:

haconf -makerw 		(open the config for writing)
hasys -freeze -persistent -evacuate <system>	(switch all SGs and
freeze the node)
haconf -dump -makero	(close and save the config)

The benefit of this is 1 command to switch all SGs and no matter how
many times you have to reboot because of OS patches, etc., VCS won't try
failing over to that node.  Freezing a system is very similar in concept
to freezing a SG.

Once you have finished the process, undo the freeze:

haconf -makerw
hasys -unfreeze -persistent <system>
haconf -dump -makero

Then repeat the process for the next node that needs to be patched.

Brad

------------------------------

Message: 2
Date: Mon, 5 Feb 2007 10:12:50 -0500
From: "Eric Hennessey" <eric_hennessey@symantec.com>
Subject: Re: [Veritas-ha] operations and maintenance tasks
To: "JV" <jv711@yahoo.com>, <veritas-ha@mailman.eng.auburn.edu>
Message-ID:
	
<0E4ADB09AED1074180AA12896EEFDB540ADB924D@hroxchcln1.enterprise.veritas.
com>
	
Content-Type: text/plain;	charset="US-ASCII"

The usual practice for applying patches in a VCS environment is to
simply switch (hagrp -switch) all your service groups to one node, patch
and reboot the now-inactive node, switch all the service groups to the
patched node, then patch the second node.

This way, your only application outages occur during the time it takes
to switch them from one node to another, and you can plan that for times
when it's convenient.

Eric

-----Original Message-----
From: veritas-ha-bounces@mailman.eng.auburn.edu
[mailto:veritas-ha-bounces@mailman.eng.auburn.edu] On Behalf Of JV
Sent: Monday, February 05, 2007 12:08 AM
To: veritas-ha@mailman.eng.auburn.edu
Subject: [Veritas-ha] operations and maintenance tasks

VCS folks -

operations and maintenance question - I am looking for best practices in
the field...

say you have a 2 node VCS cluster that runs a pile of Oracle databases
and random dependant applications. Now you want to do something like
apply the Sun recommnended patch cluster, so your "preferred" VCS node
will be down for a couple of hours.


Do you :

A) hastop -all -force
	and just knock VCS out with one punch. Now you have to bring the
"real services" up & down manually, hastart and probe if you just want
to slide VCS back in place without another disruption...


B) freeze the service groups
	before you issue hastop to prevent it migrating anywhere else of
you forget to do -all -force. Or do you just skip this entirely?


C) Disabling the Service Group
	Under what conditions is this used, other than preventing a
booting-up node with a higher preference from stealing the sg from other
node who's sleeping on the job?



Today I tried 'hares -offline sg' and it /switched/ to the other node
rather than going offline. That was not what I wanted! Later in the day
my problem was both nodes were down, and after reboot with neither
running /had/ I had to manual seed to get it up on a one-node cluster.
So obviously I am now breaking out the VCS simulator trying to practice
this VCS stuff so I don't look so lame! After 18 months of continuous
uptime, I can't say I've practiced my VCS skills that much.

thxs
JV711



 
________________________________________________________________________
____________
TV dinner still cooling? 
Check out "Tonight's Picks" on Yahoo! TV.
http://tv.yahoo.com/
_______________________________________________
Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha




------------------------------

_______________________________________________
Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha


End of Veritas-ha Digest, Vol 10, Issue 3
*****************************************

_______________________________________________
Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha
[prev in list] [next in list] [prev in thread] [next in thread]