CU Update to Sql Server 2014 hangs for 7 hours

dba.stackexchange https://dba.stackexchange.com/questions/172191

Question

How to improve time for patching installs? Applying patches on test Hyper-V VM's prior to production patching. It is a two node Always-On AG.

From Detail.txt when I let it run:

(01) 2017-04-26 15:55:36 Slp: Running discovery on remote machine: SQLHADEV1

(01) 2017-04-26 21:48:32 Slp: Discovery on SQLHADEV1 is complete

Things I've checked/done:

  1. Resources are lightly used shown in task manager
  2. Validated the Cluster
  3. Overkill on security, the account running the SQL Services on both nodes is sysadmin on both nodes

  4. Attempted to get cluster log but it error-ed

    PS SQLSERVER:\> Get-ClusterLog -Destination C:\tempsql 
Get-ClusterLog : The network name cannot be found.

  5. Tested install media on non-AG local box and it installed as expected

  6. Applied windows updates on both servers, failed over, failed back, and restarted servers. Successful automatic failover

message during stall

  1. HKLM\System\CurrentControlSet\Services\LanmanServer\Parameters\AutoShareServer to 1. restart the server service. Validate my navigating to \servername\C$. If you can't navigate there, the Admin$ share is not available.
  2. Ran the patch install on a non-AG Hyper-V VM successfully
  3. No errors or messages in the Windows Event Logs or SQL Server event logs on either node that are telling.
Was it helpful?

Solution

Changing the reg key HKLM\System\CurrentControlSet\Services\LanmanServer\Parameters\AutoShareServer to 1. restart the server service. Validate by navigating to \servername\C$. If you can't navigate there, the Admin$ share is not available. In my case something is causing the key to get reverted very quickly. I'll get with the network admin to suspend this action so the Admin$ share stays available.

enter image description here

OTHER TIPS

If you want to read the cluster log you need to do the following from powershell: import-module failoverclusters; Get-ClusterLog -Node %COMPUTERNAME%

Add an optional -Timespan integer to go back a specified number of minutes.

Then read the %WINDIR%\cluster\reports\cluster.log in Notepad.

I've looked at this file and it never ceases to make my head spin. Not particularly human readable.

I'd be interested to know what SQL server is doing during this time. What does it say it's waiting on?

Licensed under: CC-BY-SA with attribution
Not affiliated with dba.stackexchange
scroll top