When creating 128 Volumes, it failed running out of index chunks, while the total disk space is 2+ TB

[yamingk]

Trying to capture the two new issues and share with the team:
after successfully created 32 volumes and destroyed them with no problem, shutdown took more than 3 mins to finish. @Yaming Kuang is going to take a look at it.
When creating 128 volumes on a 2TB drive, we are seeing currently index vdev took 102 chunks and it failed at creating volumes because we run out of index chunks, give the current design that we preallocate index chunks during volume creation.

Possible solutions to issue #2:
Still keep the preallocation design, investigate how many index/btree ratio we need to increase and change that ratio only.
Change the design to not preallocating index chunks, but to allocate it from free pool (shared by all volume index) dynamically.