Can anybody explain what the results of vmkfstools -t 0 mean exactly ?
Background: Update ESXi 4.1 to 5.5 produced corruption on a VMFS 3 datastore.
ESXi 5.5 does not read some flat.vmdks at all and simply claims I/O errors.
vmfs-fuse allows an extraction via ddrescue but the results are vmdks with large blank areas.
An ESXi 4.1 LiveCD that I used can print the vmdk-to-disk mapping so I thought I could get better results by extracting the referenced fragments one by one.
To my surprise I got results like this:
/vmfs/volumes/489307a4-ee92425a-df7f-001f29e8362e/name # vmkfstools -t 0 namervmdk
Mapping for file name.vmdk (42949672960 bytes in size):
[ 0: 23068672] --> [Backing device/extents are offline]
[ 23068672: 8388608] --> [NOMP -- 
0 --> 8388608)]
[ 31457280: 5242880] --> [Backing device/extents are offline]
[ 36700160: 68157440] --> [NOMP -- 0 --> 68157440)]
[ 104857600: 3300917248] --> [Backing device/extents are offline]
[ 3405774848: 131072000] --> [NOMP -- 0 --> 131072000)]
[ 3536846848: 15960375296] --> [Backing device/extents are offline]
[ 19497222144: 1048576] --> [VMFS -- LVID:489307a4-abdf9df4-f385-001f29e8362e/489307a3-a08d4b9d-609d-001f29e8362e/1:( 150458341888 --> 150459390464)]
[ 19498270720: 1048576] --> [Backing device/extents are offline]
As far as I know and as far as the customer knows this datastore never has been expanded with extends.
So do I have to assume that the datastore has been expanded before the upgrade or do I read the error "Backing device/extents are offline" as a indication for corrupt entries in the VMFS pointers ?
Is there an easy way to verify wether a datastore just used a single extent or multiple ones ?
Thanks Ulli