06-06-2015 02:25 AM
Hello all,
I have a 3 to 1 Building Block setup (3 live, 1 standby)
When failing over, the 3 prod aliasses are pointed to the 1 dr-server.
Is it correct that only on 1 prod-server the services are removed? The failover succeeded, but on prod2 and prod3 I still see the indexing/storage/shopping service. USL went correct.
Everything works fine, but just checking. we'll go back tomorrow,
Does anyone know if that is as expected? I'm leaning to a yes.
Solved! Go to Solution.
06-08-2015 08:01 AM
No, I don't believe they should.
This is aimed at DR, for when those servers fail so it would be more realistic to pull the network or power cable out to simulate failed servers.
Then there's no chance they will get updated......and no potential for conflicts if you pretend one isn't working when it is.
So I don't think there's anything to worry about, but I am curious as to how prod 1 got updated. If it was a DNS cache thing then I don't think prod 1 would have been changed at all.
Andrew
06-08-2015 05:27 AM
I'm surprised it's updated that one server if the aliases are all pointing to something else.
As far as I'm aware, it uses the alias to connect to the relevant service manager, so if the aliases resolve to something else (i.e. the DR server), I'm not sure how the prod server gets updated.
Which services were affected?
06-08-2015 06:01 AM
Hi Andrew,
I assume this is due to having 3 prodservers going to 1 drserver.
Prod1, prod2, prod3, dr1
prod1 totally failed over to dr1 (on prod1, only admin/directory/taskcontroller service remained, on dr1 indexing, storage and shopping were created)
prod2 and prod3 tasks were failed over to dr1 correct (removed from prod2/3, appeared on dr1)
prod2/3 had indexing/storage/shopping service still in services, not running, but set to automatic.
There were no issues with EV functionality. We kept EV in backupmode, and tasks in reportmode, so no archiving occurred. Using EVS, restoring and searching worked fine.
I was just wondering if on prod2/3 the 'obsolete' services should have been removed too.
06-08-2015 08:01 AM
No, I don't believe they should.
This is aimed at DR, for when those servers fail so it would be more realistic to pull the network or power cable out to simulate failed servers.
Then there's no chance they will get updated......and no potential for conflicts if you pretend one isn't working when it is.
So I don't think there's anything to worry about, but I am curious as to how prod 1 got updated. If it was a DNS cache thing then I don't think prod 1 would have been changed at all.
Andrew