Guys we need some advice on sequence and best practice to implement CU on production and DR farm
Current State :
SharePoint Production Farm :
Data Center – Location A
2 WFE
3 APP
2 SQL Server Cluster ( Active / Active )
SharePoint DR Farm :
Data Center – Location B
2 WFE
2 APP
2 SQL Server Cluster ( Active / Active )
Content and MMS DBs are log shipped from Production to DR and UPS is replicated through UPS replication engine
With the above Setup we need to apply CU on Production and DR .
Assumptions / Statements :
Down time is only allowed for weekend
Data volume is around 600GB
Cant apply CU when DBs are in recovery/read mode.
Current Planned Steps for Applying CU:
Step 1.Take Full SQL backup and Pause Log shipping on Production . ( Friday Evening )
Step 2- Install CU on prod ( which may have done some schema changes and those are going into the queue ) ( Saturday )
Step 3 smoke test ( which may also generate the changes and those are going into the queue ). ( Saturday )
Step 4 User Confidence period where 1/2 day they work BAU and if everything is working as expected goto step 5 ( which may also generate the changes and those are going into the queue ). ( Monday)
Step 5 Change the DR DBs to be not in recovery mode and Install CU on DR ( Monday ) – This step will also apply schema changes to DR DBs
Step 6 Change the DR DBs to be in recovery mode and Install CU on DR ( Monday )
Step 7 Apply the paused log shipped transaction the DR DBs ( Monday )
On the Step 7 log shipping has the transaction for schema changes which is already been applied by CU installation . Can any one advise if those changes will be ignored and rest of the changes will be applied to DR DBs.
We can have another option is to detach the content db from production and apply CU and let the logshipping apply the changes to Content DBs and then re-attach the Content DBs to DR .
Can you please advise us on how this normally done
I would expect you will have problem in step 7 like you describe. once you modify the DR databases, the log shipping is no longer truly valid and it will most likely fail when trying to make the modifications. I realize that the size of the database is a little problem, but I would start down the route of:
1. upgrade prod farm2. stop logshipping3. dismount/detach content (and service DB’s) on DR farm (leaving config DB)4. apply CU to DR farm (really just laying down new DLL’s and modifying the config DB)5. Take full backup of prod DB’s and restore to DR farm.6, “reset” log shipping to start at that point.
Like I said, not extremely elegant, but I think it is the best stable option. Outage is only during step1 (of course, realize you have no DR farm to fail to until step 6 is done, if you have a strange situation happen and prod fails during that time, you have a problem to address.)
I would love to hear others opinions of this also.
