Problem description:
Under certain circumstances applications and utilities may fail with one (or more) of the following symptoms
- Event ID 157 - Disk # has been surprise removed. The original documentation for this message appears to be this blog
- Event ID 153 - The IO operation at logical block address xxxxxxxx for Disk # was retried. The original documentation for this message appears to be this blog
Which particular circumstances?
- Virtual machine
- The underlying virtual disk is storage tiered and/or the underlying file system is thinly provisioned or data deduplicated. Use of these features may cause considerable background activity, leading to I/O request completion being delayed
How can the issue be addressed?
- Increase the registry DWORD setting at HKLM\System\CurrentControlSet\Services\Disk\TimeoutValue. The default value is 60
- It should be noted that this blog entry recommends reducing the value to reduce side-effects. It should also be noted that the blog was written way before large-scale virtualisation, storage-tiering, thin provisioning or data deduplication
What would I recommend
- It was noted that the Azure Windows Server 2012 R2 OS images had 120 set as a default
- A value of 180 does not appear to have any unexpected side-effects in my environment (Windows Storage Spaces that are storage-tiered and thinly provisioned and data deduplicated)