KB3020116 - FIX: "Non-yielding scheduler" error occurs and AlwaysOn Availability Group transits to RESOLVING state

Symptoms

Assume that you have an AlwaysOn Availability Group that is deployed across server S1 and server S2 in Microsoft SQL Server 2014. A health issue is detected on the primary replica (S1), and the availability group transits to the RESOLVING state and begins failover if it is configured for automatic failover.
The availability group may remain in the RESOLVING state. The non-yielding scheduler error may appear in the error log at the primary replica (S1) or the secondary replica (S2):

• The following non-yielding scheduler error may occur on the primary replica around the time when the availability group transits from PRIMARY to RESOLVING:

  <Date> <Time> spid<ID>     Using 'dbghelp.dll' version '4.0.5'
  <Date> <Time> spid<ID>     Using 'dbghelp.dll' version '4.0.5'
  <Date> <Time> Server      Using 'dbghelp.dll' version '4.0.5'
  <Date> <Time> Server      ***Unable to get thread context for spid 0
  <Date> <Time> Server      * *******************************************************************************
  <Date> <Time> Server      *
  <Date> <Time> Server      * BEGIN STACK DUMP:
  <Date> <Time> Server      *   <Date> <Time> spid<ID>
  <Date> <Time> Server      *
  <Date> <Time> Server      * Non-yielding Scheduler
  <Date> <Time> Server      *
  <Date> <Time> Server      * *******************************************************************************
  <Date> <Time> Server      Stack Signature for the dump is 0x0000000000000176<Date> <Time> Server      Timeout waiting for external dump process 982676. <Date> <Time> Server      Process 0:0:0 (0x11428) Worker 0x00000075CB92C160 appears to be non-yielding on Scheduler 0. Thread creation time: 13011925023676. Approx Thread CPU Used: kernel 0 ms, user 0 ms. Process Utilization 2%. System Idle 84%. Interval: 76880 ms.

• The following non-yielding scheduler error may occur on the secondary replica if the availability group is configured for automatic failover and the failover partner is trying to transit to the PRIMARY role:

  <Date> <Time> spid<ID>     The availability group database "agname" is changing roles from "RESOLVING" to "PRIMARY" because the mirroring session or availability group failed over due to role synchronization. This is an informational message only. No user action is required.
  …
  <Date> <Time> Server      Using 'dbghelp.dll' version '4.0.5'
  <Date> <Time> Server      ***Unable to get thread context for spid 0
  <Date> <Time> Server      * *******************************************************************************
  <Date> <Time> Server      *
  <Date> <Time> Server      * BEGIN STACK DUMP:
  <Date> <Time> Server      *   <Date> <Time> spid<ID>
  <Date> <Time> Server      * Private server build.
  <Date> <Time> Server      *
  <Date> <Time> Server      * Non-yielding Scheduler
  <Date> <Time> Server      *
  <Date> <Time> Server      * *******************************************************************************
  <Date> <Time> Server      Stack Signature for the dump is 0x000000000000006D
  <Date> <Time> Server      External dump process return code 0x20000001.
  External dump process returned no errors.
  <Date> <Time> Server      Process 0:0:0 (0x1e94) Worker 0x000000082F270160 appears to be non-yielding on Scheduler 0. Thread creation time: 13059453624681. Approx Thread CPU Used: kernel 0 ms, user 0 ms. Process Utilization 3%. System Idle 84%. Interval: 70358 ms.
  <Date> <Time> Server Process 0:0:0 (0x998) Worker 0x00000000B3F86160 appears to be non-yielding on Scheduler 2. Thread creation time: 13059458965740. Approx Thread CPU Used: kernel 0 ms, user 0 ms. Process Utilization 3%. System Idle 83%. Interval: 76913 ms.
  Date> <Time> Server      Process 0:0:0 (0x1a64) Worker 0x0000000B5E220160 appears to be non-yielding on Scheduler 3. Thread creation time: 13059466511951. Approx Thread CPU Used: kernel 0 ms, user 0 ms. Process Utilization 3%. System Idle 83%. Interval: 76944 ms.

Note This issue also occurs in SQL Server 2012.

Resolution

After you apply this hotfix, the non-yielding scheduler condition can be avoided.
This issue was first fixed in the following cumulative update of SQL Server. 

Status

Microsoft has confirmed that this is a problem in the Microsoft products that are listed in the "Applies to" section.