SQL AlwaysOn Availability Group modifies preferred owner node on move/failover

i have 2 node windows server 2012 r2 failover cluster 2 sql server 2012 enterprise alwayson availability groups configured. within cluster have configured preference order on node alwayson availability group should run.

get-clustergroup -name "alwayson1" | set-clusterownernode -owners "node1","node2";  get-clustergroup -name "alwayson2" | set-clusterownernode -owners "node2","node1";

after configuring preference , retrieving preference order configured should be. when failover preference order adjusted first node runs on , second other node instead of leaving preference order was. doesn't matter if use powershell "move-clustergroup" or "alter availability group [alwayson1] failover" t-sql command. cluster log shown preference modified after resource has been moved.

info  [gum] node 2: executing request locally, gumid:15859, action: /rcm/gum/setgrouppreferredowners, # of updates: 1
info  [rcm] rcm::rcmgum::setgrouppreferredowners(alwayson1,<vector len='2'>

the default groups within cluster "available storage" , "cluster group" keep preference setting when configured (only configured testing verify has sql server availability group resource type).

the windows server 2012 r2 servers updated using latest windows updates , sql 2012 has build number 11.0.5556.

for other cluster groups have configured in past file, hyper-v (used affinity vm's have lot of traffic between them within same virtual switch reduce latency , additional hops. nice if vnext have affinity option cluster groups besides anti-affinity within current version)

i know if explain why sql server availability group resource modifies group preference. might understand 3 node cluster first 2 running in synchronous commit mode , third 1 in asynchronous commit mode.

thanks in advance,
dennis van den akker

hi dennis van den akker,

do not use failover cluster manager manipulate availability groups, example:

do not add or remove resources in clustered service (resource group) availability group.

do not change availability group properties, such possible owners , preferred owners. these properties set automatically availability group.

do not use failover cluster manager move availability groups different nodes or fail on availability groups. failover cluster manager not aware of synchronization status of availability replicas, , doing can lead extended downtime. must use transact-sql or sql server management studio.

the related kb:
failover clustering , alwayson availability groups (sql server)
https://msdn.microsoft.com/en-us/library/ff929171.aspx

the similar thread:
sql server clustering 2008 (multi node)

https://social.msdn.microsoft.com/forums/sqlserver/zh-cn/c35fe863-3fed-4a6e-830b-b79aa032c198/sql-server-clustering-2008-multi-node?forum=sqldisasterrecovery

best regards,

---

please remember mark replies answers if , unmark them if provide no help. if have feedback technet support, contact tnmff@microsoft.com

Windows Server  >  High Availability (Clustering)