Link to home
Create AccountLog in
Avatar of aethion
aethion

asked on

What happens to Exchange 2010 DAG when one of three nodes goes offline?

We have a three node DAG.  Becuase of the odd number of nodes the DAG does not use the file share witness that I did define on a CAS/HUB server.  I noticed when one node goes down (reboot, etc.) the cluster fails due to there not being a majority node set.   The active databases on the node I reboot to simulate a failure do not automatically transfer to the other node where the passive DB is.  

Is this normal behavior?  I was hoping that, when one of the nodes went down the DAG would use the FSW and continue to be operational/healthy.   Any suggestions on configuration based on practical experience would be appreciated.  Thanks!
ASKER CERTIFIED SOLUTION
Avatar of Akhater
Akhater
Flag of Lebanon image

Link to home
membership
Create an account to see this answer
Signing up is free. No credit card required.
Create Account
Avatar of aethion
aethion

ASKER

OK, I will have to keep testing.   Are the active DBs supposed to auto fail over to another node when one node goes  down?   Or will it always require a manual activation?  Thanks.
by default if you have not changed anything it should be automatic.

the 3 servers are in the same location ? when all 3 are up do you have any errors in the event log ?

when you shutdown one of the nodes what becomes the status of the databases?
Avatar of aethion

ASKER

All three nodes are in the same site.   No errors in event log, all else seems to be working well.  
If I shutdown one node, the DB status on active DB changes to "disconnected"  and the passive DB doesn't seem to come online (I think it said "suspended")   I'm not able to confirm this right now, I can later.  

What should the failover time be if an active DB's node goes down?  I think I had waited maybe 3-4 minutes.  Maybe there is a heartbeat setting, etc. to adjust how soon the passive node comes online?  Thanks again.
Before shutting down the node are all databases healthy ?

by "active DB" you mean the node you shutdown ? if so it should be "ServiceDown"

By passive DB you mean on another node then the one you shut down ?

failovertime is just a few seconds <30sec.

Avatar of aethion

ASKER

yes, one says "mounted" the other "healthy."    you are correct, it said "service down" and not disconnected.    I will continue to test, just wanted to confirm what expected behavior should be.  thank you.
so you have 2 copies of your database ? what about the 3rd server ?
Avatar of aethion

ASKER

right, only two, at least for now.  third server I will use to host the archive databases that I want passive copies on the other two nodes.  at least that is my initial though, maybe could be better thought out in terms of redundancy...  I know the archive DBs dont' have to live in a DAG but I still want the protection for all data.
And when the primary goes ServiceDown what becomes the status of the passive ? Healthy and disconnected ?
Avatar of aethion

ASKER

exactly (that's where I saw disconnected)  sorry I'm not in front of those systems today.
SOLUTION
Link to home
membership
Create an account to see this answer
Signing up is free. No credit card required.
Create Account