EV 12.0.3 with SQL 2012R2 Availability Groups

[SYM-AJ]

I have a new installation of EV 12.0.3, running on a SQL backend of 2012R2 utilising Availability Groups.

All seems to function OK, although I was noticing some significant 'slow responses' to regular Admin Console work, but these were sporadic.

After some initial testing of Exchange Mailbox arhiving I was noticing that either response was good, otherwise recalls were timing out and simply returning the shortcut content.  As I say, the 'issue' seems to come and go.  The odd time when starting the VAC i got a message saying the Directory Database was unavailable or directory service was not started (it was).

Initially I ran the installation, then moved the databases to the Availability Group and ran the powershell script to tell EV about the AG.  When I now run the script Get-EVDatabaseDetail it shows all the databases are in the AG and available.

My feeling is that we are having intermittent connectivity issues, possibly between the EV Server (single server) and the SQL databases.

We have 2 DNS entries for the SQL listener name - one for each of the IP addresses pn each SQL node.

Has anyone had any issues with EV12 and SQL AG's ?  Do we need to 'tweak' any other parameters or DSN's ?

AJ

[Peter_Oldenburg]

Hi,

As far as I know AG is not supported on EV12.

Further information https://download.veritas.com/resources/content/live/SFDC/98000/000097605/en_US/compatibility_charts....

Page 14: e: AlwaysOn Availability Groups are not supported on SQL Server 2012 SP2 CU3,CU4, and 2014 CU5

[SYM-AJ]

Peter,

It is not supported with those specific Cumulative Updates, but is supported with SP2 alone - which is where we are at.  This information has been confirmed by support.

It appears that EV (certainly at 12.0.3) is not supporting the MultiSubNetFailover parameter, so to get around this issue we have removed the listerner from the AG, remved the two DNS entries for the listener (both had the listener name but one pointing to each SQL node), added a DNS Alias CName entry with the original listener name pointing to the SQL node hosting the primary databases.  IN the event of an AG failover we will need to amend the DNS entry to point to the other SQL node.  Not ideal, but until support can come up with something better we have no choice.  

AJ