November 25, 2011 at 8:34 am
The cluster is sql 2005 Enterprise 64-bit running on Windows Server 2008R2 64-bit. Since the new node is active and does not yet have SP4 some things don't work right with the resource DB at SP4 and master now apparently at RTM. If we fail back to another node, add SP4 to the new node, will the shared master database on the SAN get updated to SP4 even though its applied to a node that isn't active?
November 25, 2011 at 8:53 am
Apply Service pack to the newly added node. It will only update binaries installed on that node.
The databases have already been upgraded by your previous SP install.
-- Gianluca Sartori
November 25, 2011 at 9:19 am
The problem is that the master database appears to be back at RTM but the resource database shows SP4. Both of those are on a shared storage drive -- the cluster nodes don't have any local databases.
November 25, 2011 at 10:05 am
Fail the instance over to the node that is not upgraded - shut down the passive node and then apply the service pack to that node.
If you don't shut down the passive node, the service pack will fail because it will check the other node which is already upgraded.
This only occurs with 2005 - because the way the service packs are applied for 2008 has been changed. You could try to use the /passive switch on the command line - and that might avoid the check of the other node.
Jeffrey Williams
“We are all faced with a series of great opportunities brilliantly disguised as impossible situations.”
― Charles R. Swindoll
How to post questions to get better answers faster
Managing Transaction Logs
November 25, 2011 at 10:12 am
Jeffrey Williams 3188 (11/25/2011)
Fail the instance over to the node that is not upgraded - shut down the passive node and then apply the service pack to that node.If you don't shut down the passive node, the service pack will fail because it will check the other node which is already upgraded.
This only occurs with 2005 - because the way the service packs are applied for 2008 has been changed. You could try to use the /passive switch on the command line - and that might avoid the check of the other node.
Nice tip, thank you.
-- Gianluca Sartori
November 25, 2011 at 10:18 am
by shut down the passive node I believe jeffrey means pause the node via cluster manager
---------------------------------------------------------------------
November 25, 2011 at 10:23 am
Thanks !!!
November 25, 2011 at 10:42 am
george sibbald (11/25/2011)
by shut down the passive node I believe jeffrey means pause the node via cluster manager
Nope, means exactly what I stated. Shut down the node - power off...
Don't know if pausing the node via cluster manager will work - because the way the installer for 2005 works. The 2005 installer is cluster aware - so it looks to the cluster for all available nodes where that instance is installed and attempts to perform the installation on each node.
Because one of the nodes is already upgraded - the installation fails. If the node is available, but paused I think the installer will fail because it cannot get to that node. But, I could be wrong - like I said I have never tried that.
Jeffrey Williams
“We are all faced with a series of great opportunities brilliantly disguised as impossible situations.”
― Charles R. Swindoll
How to post questions to get better answers faster
Managing Transaction Logs
November 25, 2011 at 10:51 am
pausing the node was how I last applied a patch to a multi site cluster, see here
Its a workaround for when a SQL 2005 patch fails to apply to all nodes by the standard method.
I would think it would work in this instance
---------------------------------------------------------------------
November 25, 2011 at 11:59 am
Well, I don't know if that would work or not - seems like it would, but you are also referencing a geo-cluster which is a bit different.
Either way, to upgrade the other node you have to have SQL Server active on that node and the other node inaccessible so the installer will not try to upgrade that node.
Jeffrey Williams
“We are all faced with a series of great opportunities brilliantly disguised as impossible situations.”
― Charles R. Swindoll
How to post questions to get better answers faster
Managing Transaction Logs
November 25, 2011 at 12:49 pm
I'm aware of the process being used on standard clusters
of course another option might be to remove or evict the new node from the cluster and re-do it correctly.....:-)
---------------------------------------------------------------------
Viewing 11 posts - 1 through 10 (of 10 total)
You must be logged in to reply to this topic. Login to reply