Does anyone have any downtime estimates when performing the upgrade from EV 10.0.4 to EV 11.0.1? I understand this will be site specific depending on how large the databases are, but I am just looking for some numbers from people that have already performed the upgrade.
Solved! Go to Solution.
It depends a LOT on your databases. If you do not have maintenance on them (reindexing and all that), and have lots of items (millions) in the WatchFile tables, you can be looking at days...
A real world example:
Upgrading EV on 4 servers (1 site) with directory db of 700MB, and largest Vault Store database being close to 18 GB, the upgrade from 10.04 to 11.0.1 AND CHF1 took 3 hours. We calculated 4, but finished ahead of time due to the evdatabaseupgrader.exe. I also did some small sites (2 servers) took 2 hours. I do do several checks in between, just to be sure.
But! Do you checking before starting to upgrade. Make sure everything is backed up properly, and processed from the databases. Set the archiving tasks and provisioning in report mode, clear the site schedule, restart all services. Set backupmode on stores and indexes. Make sure your MSMQ is empty. Make sure you have enough space on SQL to perform the upgrade. Install and Run deploymentscanner ahead of the actual upgrade day. Install .Net 4.5 and the necessary features before hand. Make sure your databases and SQL are having had their recommended maintenance, and make sure you have a DBA at hand to help.
Now, before I scare you off :)
The upgrade is really simple. I assume you know:
Stop all EV-services, disable the EV Admin service.
Backup SQL databases
Install EV 11.0.1 on all EV-servers.
When done (can take a while due to registering files etc) on all 4, reboot ONLY 1 of them. When rebooted, verify services are still stopped. When so, execute the EVDatabaseUpgrader. When that completes succesfully, enable admin-service, start directory service. Verify Console opens. Do not do anything else.
Reboot the other EV-servers. Enable admin service on those, start directory service on those. When all can open console, start other services on all ev servers, verify eventlog!
When ok, either leave running for a day (just in case), or be brave (like me :) ) and enable the schedule, set tasks in normal mode, and clear backup mode.
IF you want to apply CHF1 right after that. Same procedure. I did not make a backup of the db's, as there ' should' be a rollback possible. Follow documentation on that one. Then decide of you want all of your users to have the new search, or just a few (POC).
What Gert said
SQL is always the potential bottleneck so make sure you have no SQL Index fragmentation in the tables, DBCC can give you this information, and ensure that there as few items as possible in the Awaiting Backup and Awaiting to Index. You can see this from the VAC and Status view of possible issues