Pergunta

This is using RC0.

I have set up a SQL Server 2012 AlwaysOn (or more accurately, highly-likely-but-not-always-on) Availability Group.

I take the Primary offline.

  1. The Secondary takes up to 25 seconds to assume the Primary role.
  2. Any queries that were executing at the time of failover returns an error to the client.

Can anyone who has tested this new AlwaysOn feature validate my results?

  1. How soon should I expect the failover to execute and complete?
  2. Does an executing query get resubmitted against the new Primary, or am I correct that it causes a "connection failure" type error to the client?
  3. Is the following form of connection string correct, i.e. no "Failover_Partner" (because this is not mirroring)?

ConnectionString=Provider=SQLOLEDB.1;Server=10.10.10.10;User ID=x;Password=y;Persist Security Info=True;Initial Catalog=z

Note: I tried the below, but all I get is "connectivity link failure" and it never recovers. One would have thought the Native client would work better?

ConnectionString=Provider=SQLNCLI11.1;Server=tcp:10.10.10.10,8888;User ID=x;Password=y;Persist Security Info=True;Initial Catalog=z

Foi útil?

Solução

  1. It depends on how long it takes the clustering to fail over and then for crash recovery to run.

  2. No, not unless you program your app to resubmit the query.

  3. Yes, provided that you setup an AlwaysOn listener (that might not be the correct name) and the IP 10.10.10.10 is the IP address which the cluster service and AlwaysOn are using for the floating listener. If that IP is an IP which is assigned to one of the specific servers which is hosting a copy of the database then no.

Outras dicas

  1. There are a couple of bugs with windows 2008 r2 & 2012 AV groups. One being, the failover taking longer than expected. A hotfix has been released

http://support.microsoft.com/kb/2687741/en-us

Licenciado em: CC-BY-SA com atribuição
Não afiliado a dba.stackexchange
scroll top