Unsolved
This post is more than 5 years old
1 Rookie
•
44 Posts
0
1244
August 24th, 2015 02:00
Recoverpoint LUN Consumed capacity mismatch
Dear Admins,
This is a very weird issue am having with my RPA's. I have the two site replication of VNX arrays through Recoverpoint 4.1 SP1 P1. When I compare the LUN consumed capacity in GBs of my production thin luns with the DR side thin LUNs, there is a very biiiiig difference in lun sizes.
Production actual capacity: 4 TB
Production consumed capacity: 600 GB
DR actual capacity: 4 TB
DR consumed capacity: 2 TB
The journal lag is showing 200 GB for which I increased the journal volumes to the recommended value. But journal lag is a different thing. Is this something related to the SPACE RESERVATION on the DR luns including the data being copied from the DR journal.
Right now the consistency group status is ACTIVE AND DISTRIBUTING. This was a similar case with other CG's which too had this issue of inc-correct capacity consumptions on DR side.
I have opened a case with EMC and they were not really helpful with this condition.
Please help.
virtualphoton
1 Rookie
•
44 Posts
0
August 24th, 2015 02:00
Adding further to this, I deleted the DR LUNs earlier to initiate a fresh full sync from the production to DR but the issue has started again. Am afraid if this is the situation, then we will run out of space soon.
virtualphoton
1 Rookie
•
44 Posts
0
August 24th, 2015 05:00
Found the issue:
This is due to the Thick Eager Zero datastores we have on the production site. This resulted in un-even data copy and reservations on the DR side.
Why would Recoverpoint have this dependency ? Can this be corrected in future releases please?
Idan
675 Posts
0
August 24th, 2015 08:00
Hi there,
RecoverPoint doesn't have any dependency here. It seems that what's going on here is that you're fully allocating zeroes on VMDK creation (which is the behavior of THICK eager zero), data is overwritten and RP is replicating that and writing it to the target LUN.
If you would use THIN datastores, you can leverage UNMAP VAAI primitive which RP supports with VNX splitter on production and replica.
Hope that helps,
Idan Kentor
RecoverPoint Corporate Systems Engineering
idan.kentor@emc.com
panks2020
6 Posts
0
October 19th, 2015 03:00
You can simply migrate the DR lun to a new lun within a pool and get the actual size same as production lun.
panks2020
6 Posts
0
October 19th, 2015 04:00
Hi there,
RecoverPoint doesn't have any dependency here. It seems that what's going on here is that you're fully allocating zeroes on VMDK creation (which is the behavior of THICK eager zero), data is overwritten and RP is replicating that and writing it to the target LUN.
If you would use THIN datastores, you can leverage UNMAP VAAI primitive which RP supports with VNX splitter on production and replica.
Can we make changes while VMDK createing (i.e would not fully allocating zero) so that Thick DS lun size would same as DR lun.
Also, if we migrate DR lun to a new lun the size comes up same as production lun. Does RPA pop up lun size waring alerts on dashboard.?