2011-11-22 06:00 AM
IHAC that use our FlexClone technology to create test environments.
Also he is using VSM to replicate it to DRP Site.
The problem is that every "snapmirror resync" command cause deswizzling process to begin. Until the process is done it is not possible to make another SM relation for other FlexClone Volume.
The diswizzling process takes between 24 to 48 hours in my case that is not acceptable by customer since their retention of creation new FlexClones its about one per 24 hours.
There are two different environments using FlexClone that both resides on the same volume.
One, has low number of big files
Second has a big number of small files.
My questions are following:
2011-11-22 10:25 AM
#1 is a very good question... it depends on if deswizzling falls under the kahuna domain or wafl_extent... I think it is Kahuna (any wafl or snapmirror engineers who can confirm?)...and kahuna does not run on more than one core so if that is the case then it won't run much faster (a little with faster cpus, but still bound to one core)... if wafl_extent across cores than yes it would be faster but I don't think that is the case.
There is a way to stop deswizzling...that means slower access on the target since the logical to physical block reference isn't updated...so more i/o to get to blocks without the deswizzling finished but that might be an option to talk about with support... also less mirror updates (not often feasible).
You could also add more drives for better aggregate i/o which can help deswizzling too if disk i/o is the bottleneck.
2011-11-22 10:32 AM
I will try upgrading to 8.0.2P3 (if not already on a 8.X code) and check for any change in behavior.
FAS 6080 is going to be definitely faster but I am hopeful that there will be some other workaround.
2011-11-24 01:25 AM
Thank you Scott for response but desabling deswizzling isn't a way since you can make a new SM relation on flex clone volume without it.
The error the customer is getting is:
Snapmirror resynchronization of temtsh_nfs to netapp1a-snap:temtsh_nfs : Volume is a clone and scanners have not completed on the parent volume: cannot execute operation
netapp1a-snap:temtsh_nfs netapp3:temtsh_nfs Uninitialized 06:46:58 Idle
2011-11-24 01:34 AM
Since the 6080 is more powerful i also hope it will run faster.
Probably I need to be in touch with engineering team to confer it prior i am running to the customer with it :-)
2011-11-24 06:59 AM
blah...that's right. Are they using one of the retained snapshots or are they creating a snapshot for the clone on the source so that it doesn't change during the clone operations? It probably won't matter much since every update kicks off deswizzling.