Hello Everyone,
Based on my testing, incremental p2v functionality is not compatible
with CBT functionality. Please VMware to confirm this?, or provide a
way to work around it.
Here is what I'm doing:
1) Create a converter job through the VMware converter client (using
their GUI)
a) Modified the target disk to be thin
b) Scheduled a p2v synchronization activity for some time in the
future (so it doesn't run yet) and uncheck the final synchronization option
2) Run the conversion (takes about 10 minutes)
3) After the conversion is complete, create a VADP backup set for that
VM, using the default options
4) Run backup, the VM is reconfigured to use CBT, but reading active
blocks of the thin disk fails
a) This is likely because the converted VM has snapshots, and
VMware recommends removing all snapshots before enabling CBT, but
removing those snapshots will likely break the next incremental p2v job.
5) The VM is backed up as a thick disk (takes about 6 hours)
6) Remove schedule on p2v synchronization and force a synchronization
activity manually
7) VM synchronizes (takes 1 minute)
8) Run another backup session on the converted VM.
a) The CBT fails with a file fault, will not do incremental read
b) Listing active blocks fails, will do a full read of the dist as
thick (estimated time: another 6 hours)
Based on this, even though the p2v synchronization is faster, the main
bottleneck is with the VADP read, which does not speed up.
There are two issues here:
1) We need to enable CBT before the p2v creates its snapshots for the
next synch activity. Does the converter have any functionality to enable
CBT, or create a custom configuration on the target VM? If this can be
done, at least the disk will be read as a thin (maybe).
2) Why does CBT get file faults when querying changes after a p2v
synchronization? It may be fixed if CBT is enabled from the start, but
there may be other causes. Does VMware provide any guide or suggestions
for using CBT and p2v synchronization together?
Regards,
Gerson