January 30, 2015 at 5:51 am
Hi, I'm applying a service pack to 2008r2 cluster 2 node cluster with mirroring to another 2 node cluster in DR, i was planning to perform a rolling patch.....
My Question is this.... i have a large downtime window that has been agreed, so maximum availability of service is not an issue and it has been suggested to me that i should just take all the sql server services in prod and DR down and then patch them and bring them back up as this is easier than going through the rolling patch process.
Can you take SQL offline and still patch it in a cluster? If the SQL server service is offline does it matter what order you patch the nodes in the cluster or the instances with the mirrored/primary databases on? Has anyone ever done this?
I've only ever rolling patched clusters with SQL online and any advice would be appreciated.
January 30, 2015 at 6:21 pm
I would not advise to bring them all down and patch them and then bring them back up. A big reason to do a rolling patch process is to make sure everything works properly when a patch is applied. If you patch them all at once, then it will be twice as painful if you have to roll the patches back.
February 1, 2015 at 7:37 am
j.snowden (1/30/2015)
Hi, I'm applying a service pack to 2008r2 cluster 2 node cluster with mirroring to another 2 node cluster in DR, i was planning to perform a rolling patch.....My Question is this.... i have a large downtime window that has been agreed, so maximum availability of service is not an issue and it has been suggested to me that i should just take all the sql server services in prod and DR down and then patch them and bring them back up as this is easier than going through the rolling patch process.
Can you take SQL offline and still patch it in a cluster? If the SQL server service is offline does it matter what order you patch the nodes in the cluster or the instances with the mirrored/primary databases on? Has anyone ever done this?
I've only ever rolling patched clusters with SQL online and any advice would be appreciated.
Patch the cluster supporting the mirror database first. Within that cluster perform a rolling upgrade on the passive node. Once competed failover the DR instance to the passive node which will now become active. Patch the new passive node and failback the DR instance after if you prefer to.
Once the work is done in DR do the same to the live site cluster, patch the passive node first. Once this is done you'll need a small outage to failover to the newly upgraded passive. Once done complete the work on the new passive node and failback afterwards if you prefer
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply