0

This is one of those weird questions I would expect to be documented (Especially on Tech Net), but I honestly haven't found an actual explanation. As of Server 2012 R2 you can set your cluster nodes to drain on shutdown. However, as is the case sometimes, the drain procedure will not actually succeed. At such a time, it seems like the server will hang on this procedure for a long time.

What I want to know is, what is the maximum timer the server will attempt to drain all roles for, and is there any way to change this timer?

Naryna
  • 271
  • 1
  • 3
  • 13

1 Answers1

2

If the "DrainOnShutdown" property is set, the "ShutdownTimeoutInMinutes" property is effective before the cluster will terminate.

what is the maximum timer the server will attempt to drain all roles for, and is there any way to change this timer?

Yes, default is 20 minutes; more info here

There is no separate property for "drain on shutdown" vs "not drain / normal shutdown"

it seems like the server will hang on this procedure for a long time.

I suggest you check the timeout property (Get-Cluster).ShutdownTimeoutInMinutes, it is probably set to 20 minutes, which is indeed a long time to wait

Edwin van Mierlo
  • 333
  • 1
  • 9
  • 1
    Yes, this is exactly the information I was looking for. I have no idea why the article for DrainOnShutdown doesn't mention ShutdownTimeoutInMinutes. As a horrible coincidence, somehow our clusters had been set to 149 minutes instead of the default 20. 20 is bad enough, but 149 is terrible and odd. – Naryna Jun 21 '16 at 15:48
  • @BrandynBaryski please see the note on the info indicated "If the Hyper-V role is installed, the default time-out value is calculated based on the size of physical memory, using a ratio of 100 minutes / 64 GB. The minimum computed value is 20 minutes." -- as you have indicated that this is a Hyper-V install, this could be your culpritt – Edwin van Mierlo Jun 21 '16 at 15:50