'RE: [Veritas-ha] FW: Mail delivery failed: returning message to s'

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

List:       veritas-ha
Subject:    RE: [Veritas-ha] FW: Mail delivery failed: returning message to s
From:       Paolo Supino <paolo () telmap ! com>
Date:       2003-05-21 9:10:13
[Download RAW message or body]

Hi Mirko 

   I still didn't try your solution (but I'm about to). I simply have to
write you this email before 
I forget; I don't have a problem with the sqlnet resource. It closes up and
goes offline without
any problem. I have a problem with the oracle resource: It tries to do
shutdown immediate 
and if that fails goes on and calls clean that tries as a final act to do
shutdown abort. But 
because there is no network it't can't connect to the database and actually
try and shut it 
down (no matter what). I want to change the shutdown abort command in the
clean script
of the oracle agent with something that will kill all the processes, but I
have a feeling that 
clean script isn't called only when there is a network timeout. What is your
experience 
in this area? 

	Paolo 

-----Original Message-----
From: m_schlottke@gmx.net [mailto:m_schlottke@gmx.net]
Sent: Wednesday, May 21, 2003 9:25 AM
To: Paolo Supino
Cc: tcrowe@ctsinc.net; paolo@telmap.com; veritas-ha@mailman.eng.auburn.edu
Subject: RE: [Veritas-ha] FW: Mail delivery failed: returning message to s
ender

Paolo, 

the problem is, that the lsnrctlr used for starting and stopping the
database connects to the virtual resource-group ip-address and it can't
connect in
case of a network failure or a down configured ip-address! A solution for
this
problem I implemented at a customers site, was to set the offline timeout of
the sqlnet ressource type to a very lower value (30 seconds), the listener
doesn't care if it's been killed. The listener was killed by the clean
script
after 30 seconds in case of a public network failure and the database could
be brought down cleanly by vcs... It worked!

Might this be a solution for your environment ?

Regards, 

-- Mirko

> Hi 
> 
>   Indeed the mounts were hard mounts and I changed them, but the 
> problem persists and the cluster hangs instead of switching. 
> 
> 
> 
> 
> 
> TIA 
> 
> 
> 	Paolo 
> 
> 
> -----Original Message-----
> From: Thomas Crowe [mailto:tcrowe@ctsinc.net]
> Sent: Wednesday, May 21, 2003 2:45 AM
> To: Paolo Supino
> Cc: 'veritas-ha@mailman.eng.auburn.edu'
> Subject: Re: [Veritas-ha] FW: Mail delivery failed: returning message to
> sender
> 
> 
> Paolo;
> 
> A possible problem *may* be if the NFS mounts are hard mounted, when you 
> try to execute any Oracle actions that generate I/O, those I/O will hang 
> until the timeout (IIRC, that is ~15 minutes in Solaris by default).  If 
> you change the mounts to soft mounts, then Oracle should fail when the 
> network goes away and can no longer access it's datafiles.
> 
> I am not going to suggest a radical design change, but you may want to 
> consider configuring ethernet trunking since you are accessing the 
> datafile via NFS.
> 
> HTH
> 
> Paolo Supino wrote:
> 
> >Hi 
> >
> >  I received responsability for a site that consists of 2 sun e220
> >computers and a netapp filer. The main objective of the site is to 
> >run an oracle database in ha (it's not my design, so please don't 
> >offer me to change it because I can't). The oracle software is 
> >installed on both solaris machines locally while the database itself
> >is located in the netapp and is nfs mounted by the active system. 
> >I decided to test the configuration by disconnecting the network 
> >cable from the active computer and to my dismay vcs hangs while 
> >trying ot terminate the oracle instance running, and fails to switch
> >to the second machine. After 2 days of trying to solve the problem 
> >I narrowed the problem to the point that the offline and clean scripts
> >try to run sqlplus to shut down the database and hang there. 
> >  Can anyone think where the problem lies and offer a solution? 
> >
> >
> >
> >
> >
> >
> >TIA 
> >
> >	Paolo 
> >_______________________________________________
> >Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
> >http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha
> >
> >  
> >
> 
> -- 
> Thomas Crowe
> Senior Engineer / Senior Architect
> EMC Proven Professional, Master Architect
> EMC Proven Professional, Master+ Operator
> CTS Professional Services, Atlanta
> Yahoo IM:  thomas_crowe
> MS Messenger: thomas_crowe@yahoo.com
> 
> 
> 
> _______________________________________________
> Veritas-ha maillist  -  Veritas-ha@mailman.eng.auburn.edu
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-ha
> 

-- 
+++ GMX - Mail, Messaging & more  http://www.gmx.net +++
Bitte lächeln! Fotogalerie online mit GMX ohne eigene Homepage!
_______________________________________________
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] 