Unsolved
This post is more than 5 years old
265 Posts
0
695
Snap new writes
Hi
if Host trying to change data ( write to already allocated block ) snaphot will copy data to Snappool
what if host trying to make new write for a bloack that have nothing what will snapshot do
will it copy nothing 000: to snap pool which will consume capacity in snap pool
PedalHarder
465 Posts
0
December 3rd, 2012 22:00
When a source device is in an SNAP relationship to a VDEV, the first update to any track on the source device (after the SNAP is activated) will result in the pre-updated version of that track being placed in the SAVE pool. If the before and after data is the same, it is still an update to the track and will be copied to the SAVE pool. The SNAP session is just looking at the track being updated, not the contents of the update.
sauravrohilla
859 Posts
0
December 4th, 2012 19:00
Agree with Jason, TF does not know the content, it only cares about the update to the protected tracks.
regards,
saurabh Rohilla
LBM99
119 Posts
0
December 5th, 2012 10:00
It depends on what you mean by the original track has "nothing." For example, if the source is Thin and the track is unallocated then a new write will not copy anything to the VDEV / SAVE Pool. But if the source is allocated but written zeros then it will copy that over.
LBM99
119 Posts
0
December 5th, 2012 13:00
Correction, that behavior is for Clone and VP Snap copy processes. Ran a few tests with Virtual and Multi-Virtual Snap and it seems that SAVE pool space will be consumed for writes to unallocated source tracks.
koteswaran
20 Posts
0
December 9th, 2012 00:00
Yes thats right, if a write to snap occurs at unallocated extend of thin device, still a data is saved and written to the save pool, but its not going to affect the source thin device in any manner, at the same time, once the snap is destroyed, it will free up the all the space from the pool whatever space was occupied by this snap