Need to rebuild a node

[No idea what I'm doing]

Hey, guys. I have a two node SG cluster running 11iv3 and SG 11.18. I need to rebuild the passive node and I'm not sure what the best way to do this is. I won't be able to use an Ignite image of the primary node to lay down on the passive one as I don't have access to a tape drive and my Ignite server is on another subnet. So my plan is to do a cold installation from the DVD.

Once the server is back up and running, I assume I copy over my cluster.conf, cmclnodelist, cmclconfig(?) and also the entire package subdirectory. Is it just then a matter of doing a cmrunnode and it's able to start up because all of the information is contained in the cmclconfig? That seems a little bit too easy.

Thanks.

[Tim Nelson]

just because the ignite server is on another subnet should not stop you ?

dbprofile

[sahir_1]

Hi,

Once after your installation completes, (including service guard) you can go for following steps to complete your service guard functional.

copy cmnodelist, cluster.ascii, pkg.cntl files to passive node.

cmhaltcl -f
cmcheckconf -c /etc/cmcluster/cluster.ascii
cmapplyconf -c /etc/cmcluster/cluster.ascii
cmruncl -v
cmviewcl (to verify)

i hope this will help.

Thanks
Sahir

[No idea what I'm doing]

I also don't want to Ignite it because I'm not fully confident in the configuration of the other node. So I want to start from scratch with this node.

I am trying to get this done without having to halt the cluster and taken an application outage. what if I tar up /etc/cmcluster on the node I'm rebuilding and save it off somewhere, then restore it to the server after it's rebuilt? Do you think I'd be able to get away with just doing a cmrunnode then?

[RajuD]

Yes u can do online cluster reconfiguration without halting the cluster and package.

Check the Pre-requisites

11iv3 with sg 11.18+ online reconfiguration is possible.

Check the os release i think it should be more than june

[No idea what I'm doing]

No one really answered my question so I'll post what I was able to find here in case anyone runs into a similar situation.

Yes, you can tar up the /etc/cmcluster directory and lay it down on the node after it's rebuilt. I had issues with the numeration of the lan interfaces and also the cluster lock lun. I used the ioinit -f command to change persistent device file number(i.e. it needed to be /dev/disk/disk895 instead of /dev/disk/disk944) in order to get the cluster to be okay with the node's cluster lock lun.

I also had to change the numeration on the LAN devices as their numbers changed after I reinstalled the operating system. The lan4 and lan5 devices were what were configured in the cluster for the primary and heartbeat. Those interfaces became lan6 and lan7 for whatever reason after I reinstalled the o/s. Using the ioinit command, I changed them to lan4 and lan5.

After those two issues were rectified, I was able to do a cmrunnode and the node started right up and joined the cluster.

You also want to make sure you have the same ServiceGuard patches installed as the other node(s).