Yeah, there's documentation somewhere on how to create your own
scripts to run on node failover.  It's good to hear that the basic
idea seems sound, if I get a free day I'll give that a try.  I have a
feeling that we might need to re-join the domain each time, but I'll
certainly try just starting the service.




On Thu, Feb 12, 2009 at 7:31 PM, Alan M Wright <[email protected]> wrote:
> On 02/11/09 18:23, Ross wrote:
>>
>> Just out of curiosity Alan, what would be needed to start CIFS reliably on
>> a domain if we assume that the hostname, IP, and ZFS pools will be migrated
>> automatically by the Cluster software.
>>
>> You would want the CIFS service stopped for any inactive node, and then
>> when the node came active, would you just need a script to:
>>
>> - Join the domain
>> - Start the SMB/Server service
>>
>> For a purely windows network (with AD doing all the authentication), is
>> there anything there I've not thought of?
>
> Off the top of my head, you could probably do this with some
> scripts and ssh commands.
>
> If you pre-configured CIFS on each node (and joined the domain)
> and left the CIFS service disabled on nodes on which it should be
> quiescent then simply enabling the CIFS service should be enough
> to make it operational.
>
> The first domain join would create the machine trust account on
> the DC and the subsequent domain joins should behave as if you
> had pre-created the machine trust account.  When a node fails
> over, as long as you can trigger an 'svcadm enable smb/server'
> on another node [I think] it should work.
>
> Alan
>
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to