I was escalated to Level II and this issue appears to be resolved.
Looking back at my notes from last time, I uninstalled and reinstalled the VMtools on the Veeam server, since VMware provides the SCSI controller driver. That Veeam server hasn't had the "cannot get file size error" since.
David in Level II looked thru the event logs of the Veeam server currently having these errors and found some matching errors in the System log:
- Code: Select all
Log Name: System
Date: 3/25/2018 4:05:01 PM
Event ID: 11
Task Category: None
The driver detected a controller error on \Device\RaidPort0.
Still sorta point to VMware as the culprit.
This time, we decided to switch to a different SCSI controller:https://kb.vmware.com/s/article/1010398
So I migrated the Veeam server in question from the LSI_SAS SCSI controller to the Paravirtual SCSI controller. This is a resource definition in the VMware settings for the Veeam server. The switch is a little tricky as it's entirely possible to blue screen the server on startup due to a missing driver.
My feeling is that this issue was caused by a bug in VMware that Veeam started tripping with 9.5. The above solution is more of a work-around than a fix as we're simply side-stepping the bug by using a different driver.
FYI, I did test the Paravirtual driver on a couple of test boxes. The first one had no issues after installing using the above KB. The second instance was nothing but blue screens. On that instance, I finally manually downloaded the Paravirtual driver from the VMware tools download directly from VMware and just installed it. After both those experiences, I was able to get that driver into production with no problems.
The Paravirtual driver documentation does say that it's a more efficient driver (less CPU, more throughput). Considering the amount of data moved by the Veeam server, that could be a win too...