Re: 10g RAC network failure behavior

Hello,

I recently I had a issue where I was testing the node resources like VIP, GSD and listener. I stopped the node resources and a side effect of this was that the related databases also went down. Because ASM went down.

This is caused by the dependency on the VIP. So if there is a VIP problem everything where there is a dependency on the VIP will fail. This is untill 10.2.0.3 normal behaviour, altough I believe this is something you don't want.
>From patch 10.2.0.3 this behaviour will change and the database/asm
will not go down if there is a vip issue.

If you use crs_stat -p <resourcename> you will see the dependencies parameter. the resource listed here are required before the resource can be started.
To change this checkout the 10.2 manual it will explain how to update the profile.

Hope this helps.

Greetings,
BCB
www.rachelp.nl

Keith wrote:
> DA Morgan wrote:
> > Keith wrote:
> > > Metalink note 296878.1 describes the behavior starting with 10.1.0.4.
> > >
> > > DA Morgan wrote:
> > >> Keith wrote:
> > >>> Well, I just inherited a 2-node 10g RAC database on linux. I'm trying
> > >>> to understand the network failure points, and what the behavior is
> > >>> supposed to be from RAC.
> > >>>
> > >>> I just read a Metalink note stating that the DB and ASM are dependant
> > >>> on the VIP, so a link or adapter failure on the public interconnect
> > >>> would cause the instance to crash. Unfortunately there is no more
> > >>> information as to why they are dependant on the the public
> > >>> interconnect. Why crash the instance? What about locally connected
> > >>> apps? Isn't there another hearbeat through the private interconnect?
> > >>>
> > >>> Also, what I haven't found is what happens when the private
> > >>> interconnect fails? CSS is supposed to guard against the split brain
> > >>> syndrome, but how? What happens?
> > >>>
> > >>> Any help would be apprecated from you RAC gurus.... and if you could
> > >>> point me to the right documentation, that would be helpful too. I've
> > >>> been perusing the CRS/RAC install and admin guides and I'm not finding
> > >>> this information anywhere.
> > >>>
> > >>> Thanks in advance for your help.
> > >> I've not seen anything in the documentation indicating a crash was an
> > >> appropriate response. Node ejection and operating system reboots on the
> > >> other hand are.
> > >>
> > >> Can you provide a link to the doc you are reading.
> > >> --
> > >> Daniel A. Morgan
> > >> University of Washington
> > >> damorgan_at_x.washington.edu
> > >> (replace x with u to respond)
> > >> Puget Sound Oracle Users Group
> > >> www.psoug.org
> >
> > 1. Please do not top post.
> >
> > 2. I have not observed it with 10gR2.
> > --
> > Daniel A. Morgan
> > University of Washington
> > damorgan_at_x.washington.edu
> > (replace x with u to respond)
> > Puget Sound Oracle Users Group
> > www.psoug.org
>
> 1. Ok. Sorry.
> 2. Thanks for the feedback.
Received on Sun Sep 17 2006 - 15:40:33 CDT