Unsolved
This post is more than 5 years old
5 Posts
0
2093
Centera - is it possible to delete not empty pool?
Hi all,
I need to delete not empty pool from Centera with CentraStar 4.2.0-3881-0-0 in Basic mode. I can not delete expired clips due to third-party soft issue so the only way to free up space is pool delete. Is it possible to do it from service account?
Greeting
Tom
Gotsche_Helmut
48 Posts
0
January 5th, 2017 14:00
Hi Tom,
There is no way to delete a pool, which contains objects. Application with proper privileges has to issue (privileged) delete calls for each content address and then Garbage collection will release the space. Each delete call leaves a small metadata file behind (called reflection) and hence a pool under this CentraStar Release is never "empty" and the "delete pool" CLI command will not work. Reflections are very small but mirrored.
holgerjakob_c0722c
337 Posts
0
January 5th, 2017 22:00
Hi Tom
A non-empty pool may not be deleted. Once all data has been deleted in that pool by either deletes or privileged deletes, the reflections remain like Helmut said. When you register the pool for reflection expiry the remaning reflections can be deleted after at least 3 months. Then the pool can be deleted. As for 4.3 you'll need at least patch 4 for this to really work.
If you want assistance with the deletion, let me know on holger.jakob@informatio.com
Best regards, Holger
tomasz_rachobin
5 Posts
0
January 6th, 2017 08:00
Thank You for information. We must play with this somehow.
Tom
tomasz_rachobin
5 Posts
0
January 6th, 2017 08:00
Thank You for reply.
I would like to ask about some other issue we have. One of out targets has became totally full and there is no way to send even replicate delete to it. I tried to manually delete some clips from remote pool but this is impossible as well. Every action finish with error
Error Number: -10156
System Error: 17
Error: FP_TRANSACTION_FAILED_ERR
is there any way (except buying new nodes) to free up some space to enable running functions?
Greetings
Tom
Gotsche_Helmut
48 Posts
0
January 7th, 2017 01:00
You are right, adding physical capacity is the easiest and most sufficient way. You can also open a Service Request to solve the problem maybe the reorganization buffer can be reduced for the deletion process and then increased to the previous size again. During that phase the system is on risk not be able to perform a recovery, but if it is the D/R site, it might be feasible
tomasz_rachobin
5 Posts
0
January 9th, 2017 09:00
Thanks a lot for this suggestion.
I've changed (or disabled) regeneration buffer to 0 and made instant cleanup in pool via JCASScript. Unfortunatelly, after this operation I don't have enough space to enable buffer.
I have 2.5TB (4%) available, when I try to enable buffer I see error "The new regenerationbuffer would result in less than 5% available capacity."
I know that this 5% value is set in Centera "by design" but maybe there is the way to change it? I would rather prefer to have regeneration buffer active
Some listing below:
Config# sho cap av
Number of nodes: 16
Number of nodes with storage role: 16
Total Raw Capacity: 58 TB (100%)
Used Raw Capacity: 54 TB (92%)
Free Raw Capacity: 4.4 TB (8%)
System Buffer: 1.9 TB (4%)
Regeneration Buffer: 0 TB (0%)
Available Capacity: 2.5 TB (4%)
Total Object Count: 803 M (100%)
Used Object Count: 2 M (0%)
Free Object Count: 801 M (100%)
Config# set cap reg
Limit in disks per cube [0]: 1
Issue the command?
(yes, no) [no]: yes
Command failed: Setting capacity regenerationbuffer limit. Can not set the regenerationbuffer size to 1 disks. The new
regenerationbuffer would result in less than 5% available capacity.