Symptoms
Virtual Machines that use SEsparse virtual disk format stops responding while running specific I/O workloads. You are unable to suspend the affected virtual machine. Restarting the management agents on the ESXi host does not resolve the issue. Host reboot is required to recover the virtual machine. In the /var/log/vmkernel.log file, you see the entries similar to: vmkernel.4:2016-09-30T22:49:50.786Z cpu46:36668)VSCSI: 2590: handle 8193(vscsi0:1):Reset request on FSS handle 134428 (3 outstanding commands) from (vmm0:RUDWV-VDIFS201) vmkernel.4:2016-09-30T22:49:50.786Z cpu28:33140)VSCSI: 2868: handle 8193(vscsi0:1):Reset [Retries: 0/0] from (vmm0:RUDWV-VDIFS201) vmkernel.4:2016-09-30T22:49:50.786Z cpu28:33140)PVSCSI: 1593: ctx=0xffffe0018c781220 cdb0=0x8a on scsi0:1 completed during reset (0x8) vmkernel.4:2016-09-30T22:49:53.788Z cpu46:36668)WARNING: VSCSI: 3697: handle 8193(vscsi0:1):WaitForCIF: Issuing reset; number of CIF:3 vmkernel.4:2016-09-30T22:49:53.788Z cpu46:36668)WARNING: VSCSI: 2627: handle 8193(vscsi0:1):Ignoring double reset 2016-09-30T22:50:21.395Z cpu28:33140)VSCSI: 2868: handle 8193(vscsi0:1):Reset [Retries: 1/0] from (vmm0:RUDWV-VDIFS201 Note: The preceding log excerpts are only examples. Date, time, and environmental variables may vary depending on your environment.
Cause
The issue occurs during I/O operations of a VM with SEsparse-based snapshots. The VM might fail when running a specific type of I/O workload in a number of threads because of interlocking SEsparse metadata resources.
Resolution
This is a known issue affecting VMware ESXi 6.x. This issue is resolved in: VMware ESXi 6.0 Update 3a, available at VMware Downloads. VMware ESXi 6.5 Patch 01, available at VMware Downloads.
