1833792 Members
1732 Online
110063 Solutions
New Discussion

samba/cifs

 
p7
Frequent Advisor

samba/cifs

hi all

i have samba running on a superdome node in a serviceguard cluster. i have the failover node also running samba, if the users n user binaries are on the shared disks, can i just swing them over n samba will work fine

thx in advance
3 REPLIES 3
Devesh Pant_1
Esteemed Contributor

Re: samba/cifs

You will have to ensure that all the applications are configured though the serviceguard packages or else you will have a different instance of the application coming up. I hope the binaries etc have some kind of an association to the host/ip address that differentitates.

One Possible way to find if everything will work fine is to shut down the application on the main server and have the samba/cifs mapping on the failover server set up and bring up the application.
Check it out if that is what you want.

I would say you will have to have the IP stuff moved over to the other server as well in order for the application to accept connections.
Please provide more details and I think this can be worked out.

thanks
Devesh
Steven E. Protter
Exalted Contributor

Re: samba/cifs

If the binaries are on shared disks, all you need to do is have a package control script set up to start the application/software(whatever) when the node fails over.

Note though: You need to test this functionality, not just rely on a template script. I know you are working with big old superdomes, but if you don't test it, a thousand things can go wrong.

SEP
Steven E Protter
Owner of ISN Corporation
http://isnamerica.com
http://hpuxconsulting.com
Sponsor: http://hpux.ws
Twitter: http://twitter.com/hpuxlinux
Founder http://newdatacloud.com
omniview
Advisor

Re: samba/cifs

Pasquale,

We are doing, essentially, the same thing that you are attempting, with a few exceptions.

We currently run two fail-over packages on MC/SG: Oracle 9i and Legato Networker. Since all CIFS activity relates to Oracle on our cluster, I simply tacked the startsmb and stopsmb scripts into the Oracle startup and shutdown scripts.

I created a folder on one of Oracle's shared VGs for Samba, copied /etc/opt/samba and /var/opt/samba to this folder, then symbolicly linked back to it. Nothing is done to the binaries on the servers and they remain in tact in /opt/samba on each node.

The effect is that the MC/SG node running the Oracle 9i controls Samba (CIFS) and SWAT. It worked for me. Hope it works for you.