We recently converted some virtualbox images from vdi to vmdk using qemu-img. This all went fine and well and was accepted properly by ESXi 4.1 datastore. Starting the images and normal operations the images appeared to be normal.
However we noticed a problem. When a snapshot was taken, it shutdown the image and then deleted it? Very strange.Checking the logs nothing really jumped out at us whilst looking, but we did notice this;
Nov 12 14:53:43 Hostd: [2010-11-12 14:53:43.371 23B40B90
verbose 'vm:/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******.vmx']
No Upgrade required Nov 12 14:53:43 Hostd: [2010-11-12 14:53:43.446 23B40B90
error 'App'] Failed to get data ring: vmodl.fault.InvalidArgument Nov 12
14:53:43 Hostd: DISKLIB-VMFS :
"/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******-boot-disk-s001.vmdk"
: open successful (17) size = 21474819072, hd = 0. Type 8 Nov 12 14:53:43
Hostd: DISKLIB-VMFS : "/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******-boot-disk-s001.vmdk"
: closed.
Nov 12 14:53:43 Hostd: [2010-11-12 14:53:43.455 23B40B90
error 'App'] Failed to get data ring: vmodl.fault.InvalidArgument Nov 12
14:53:43 Hostd:
DISKLIB-VMFS :
"/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******-data-disk-s001.vmdk"
: open successful (17) size = 85899328512, hd = 0. Type 8 Nov 12 14:53:43
Hostd: DISKLIB-VMFS :
"/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******-data-disk-s001.vmdk"
: closed.
Nov 12 14:53:43 Hostd: [2010-11-12 14:53:43.462 23B40B90
verbose 'vm:/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/TKLivePROJ/TKLivePROJ.vmx']
Time to gather config: 67 (msecs) Nov 12 14:53:43 Hostd: [2010-11-12
14:53:43.498 23B40B90 error 'App'] Failed to get data ring:
vmodl.fault.InvalidArgument Nov 12 14:53:43 Hostd: [2010-11-12 14:53:43.500
23B40B90 info 'DiskLib'] DISKLIB-VMFS :
"/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******-boot-disk-s001.vmdk"
: open successful (17) size = 21474819072, hd = 0. Type 8 Nov 12 14:53:43
Hostd: DISKLIB-VMFS : "/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******-boot-disk-s001.vmdk"
: closed.
Nov 12 14:53:43 Hostd: [2010-11-12 14:53:43.501 23B40B90
error 'App'] Failed to get data ring: vmodl.fault.InvalidArgument Nov 12
14:53:43 Hostd:
DISKLIB-VMFS :
"/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******-data-disk-s001.vmdk"
: open successful (17) size = 85899328512, hd = 0. Type 8 Nov 12 14:53:43
Hostd: DISKLIB-VMFS :
"/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******-data-disk-s001.vmdk"
: closed.
Nov 12 14:53:43 Hostd: [2010-11-12 14:53:43.507 23B40B90
verbose
'vm:/vmfs/volumes/4cc965ad-0b25d3d5-795b-00269e3c9f3a/*******/*******.vmx']
Time to gather config: 11 (msecs)
It appears that vmware is losing its disks, and if you tailed the logs and then used the server this would continously appear in the logs. We ensured that the system was fully up to date, and we had some images which were on the system which were created from scratch on the ESXi software which didn't throw up errors like this.
Add to the problem even more it appear it also affects the connection to the hardware also. We saw underlying issues with performance to the hard drives. 600GB 10K SAS drives showing seq. reads of 39.2MB/sec writes of 2.4MB/sec. If the VM was built on ESXi from scratch we saw performance of seq. read of 123.4MB/sec & seq. write 67.9MB/sec (This is on RAID 1 Configuration). This is also no rubbish kit we were on as it was an a dell M610 using 2 x 6core L5640 with 64GB RAM.
This appears to be a bug somewhere within either vmware, qemu-img. Does anyone have any similar issues or possibly a way of resolving this issue?
