Issue: AlwaysOn fail over for long time

Cloud Services
,
1

Hi

we are facing the issue that it will spend about 5 min to fail over the database cluster with alwayson,

thus could be a risk to make MES service unvailable for long time when fail over happened

so could you confirm that it is normal or not?

if not, what should we do to solve it?

2

Dear Fu,

The behavior you’re describing although possible is not normal.
Usually failover on always-on clusters takes a few seconds only.

In order to troubleshoot the issue, here are some pointers where start:

  • When you notice this behavior, are there transactions being executed in the database?
    If a transaction is open to the database, the system will try to wait until the transaction is completed. If you have long-running transactions this might lead to the behavior you’re describing. This can happen when (non-exhaustive list):

  • Maintenance Jobs are running in the database;

  • The replication cycle is taking longer than normal (usually it takes 5s or less);

  • Understand what’s your witness mechanism for Always-On.
    Traditionally this is your Active Directory controller(s), although it’s possible to setup Always On without Active Directory. If you’re using Always-On with AD, please make sure you’ve a reliable and low-latency to your Active Directory controller. If you have multiple controllers, make sure your machine is connecting primarily the closest AD controller;

Hope this helps,
José

3

ok, thx. it 's faster (less 10 seconds) than before after we made some adjustments on network between AD and AlwaysOn cluster.