VMware ESX unrecoverable error: (vcpu-#) vmk: vcpu-#:Memory allocation failed for PFrame Dir
vmware.log
of the virtual machine there are messages similar to:[YYYY-MM-DDTHH:MM:SS] Wa(03) vcpu-# - A core file is available in "/vmfs/volumes/<datastore-uuid>/<vm-name>/vmx-zdump.002"
[YYYY-MM-DDTHH:MM:SS] In(05) vcpu-# - Msg_Post: Error
[YYYY-MM-DDTHH:MM:SS] In(05) vcpu-# - [msg.log.error.unrecoverable] VMware ESX unrecoverable error: (vcpu-#)
[YYYY-MM-DDTHH:MM:SS] In(05)+ vcpu-# - vmk: vcpu-0:Memory allocation failed for PFrame Dir
vmkernel.log
of the ESXi host there are messages similar to:[YYYY-MM-DDTHH:MM:SS] cpu33:6815997)WARNING: VmAnon: vm #####: 1697: kernel data alloc exceeds reservation
[YYYY-MM-DDTHH:MM:SS] cpu33:6815997)WARNING: World: vm #####: 8534: Simultaneous panic! :vmk: vcpu-0:Memory allocation failed for PFrame Dir
[YYYY-MM-DDTHH:MM:SS] cpu33:6815997)P2MCache: 427: vm #####: GetPhysMemRange failed for PPN 0x418600 canBlock 0 count 95 status Out of memory
[YYYY-MM-DDTHH:MM:SS] cpu33:6815997)PVSCSI: 956: PROCESS: failed to pin: Out of memory
VMware vSphere ESXi 7.0.x
Memory compression is enabled by default. The issue is caused by the corrupted memory compression of PVSCSI. When there are multiple HotAdd vmdks with one PVSCSI controller, it causes a memory leak and panic/crashes the virtual machine.
More information about memory compression, see Memory Compression
Currently the issue is under investigation.
As a workaround of the issue, perform either of the following steps:
sched.mem.zip.enable = FALSE
' to the virtual machine to disable memory compression as shown in Edit Configuration File Parameters