How It Works: 17888 – All Schedulers on Node Appear Deadlocked


Error 17888: All schedulers on Node ## appear deadlocked due to a large number of worker threads waiting on <<RESOURCE>>. Process Utilization ##%.

SQL Server creates a Scheduler Monitor thread for each node.   One of the Scheduler Monitor tasks is to watch the schedulers assigned to the node for progress.   Slow progress triggers additional checks; among them is the 17888 check.

 

The 17888 condition is when Scheduler Monitor detects the workers are not progressing and if this is because of a common resource bottleneck.

 

The following is an example showing clients not processing results.  This could be a bad application, router problems, or other network issue.

 

Example: All schedulers on Node 0 appear deadlocked due to a large number of worker threads waiting on ASYNC_NETWORK_IO. Process Utilization 0%.

 

The message indicates that a large number of workers are waiting on network I/O and until the I/O completes the SQL Server will remain bottlenecked.  

 

Address the <<RESOURCE>> bottleneck to resolve the problem.   The DAC connection is a great way to obtain additional information about the bottleneck.

 

Reference: How To Diagnose and Correct Errors 17883, 17884, 17887, and 17888
http://www.microsoft.com/technet/prodtechnol/sql/2005/diagandcorrecterrs.mspx

Bob Dorr
SQL Server Principal Escalation Engineer