Understanding Always-on Availability Group Session-Timeout parameter

• szejiekoh

  SSCarpal Tunnel

  Points: 4155

  More actions

  September 21, 2016 at 10:51 am

  #400755

  Hi all,

  I have a 2 node WSFC with node1 hosting a primary replica and node2 with a secondary replica.

  When going to AG -> Properties, i see that i can set the session-timeout for both the primary and secondary replica.

  My understanding of "session-timeout" is how long a primary replica waits for a response from a secondary replica.

  Hence ->

  q1) What is the meaning of setting a "session-timeout" on the secondary replica ?

  q2) If the session-timeout is mutually applicable to both the primary and secondary replica, then whats the point of having this parameter individually for each of the replica ?

  -- is there any scenario you would set 10 seconds for primary and 15 seconds for secondary ?

  q3) the last confusing bit is in the documentation ->

  https://technet.microsoft.com/en-us/library/hh213612(v=sql.110).aspx

  where group_name is the name of the availability group, instance_name is the name of the server instance that hosts the availability replica to be modified, and seconds specifies the minimum number of seconds that the replica must wait before applying log to databases when acting as a secondary replica. The default is 0 seconds, which indicates that there is no apply delay.

  -- has this parameter got anything to do with apply delay ?

  In short,

  - is this parameter only applicable to a primary replica - in which it defines how long it will wait to receive a ping message from a secondary replica ? or it is applicable to all secondary replicas as well (in that case, why can't we just have 1 common timeout setting) ?

  - or base on the quote above, this parameter when set at secondary replica defines the delay that a secondary replica will apply the changes in the log to the database ?

  This seems to be pretty confusing to me.

  Hope some gurus can shed some light.