Checkpoints not registed into the mount table

Question

Hi folks,i had a few file systems running out of SavVol space, if i would run nas_ckpt fs_name -list, it would show that older checkpoints were INACTIVE. As i understand it got to the point where there was no more space in the pool so SnapSure started deleting older snapshots. Now i have presented more space to Celerra and when i run nas_ckpt list command again ..it shows 'n' in the inuse column: 448 Checkpoint_Daily_015 04/30/2008-20:00:12-EDT y 90% 81% 452 Checkpoint_Daily_016 05/01/2008-20:00:06-EDT n 90% 81% 456 Checkpoint_Daily_017 05/02/2008-20:00:17-EDT n 90% 81% 460 Checkpoint_Daily_018 05/03/2008-20:00:18-EDT y 90% 81% 464 Checkpoint_Daily_019 05/04/2008-20:00:16-EDT y 90% 81% so what does it mean if checkpoints are not registered in the mount table ? Does it mean that if this datamover reboots, they will not be remounted ? Are they still available for restore and will the eventually get back to 'y' once they get refreshed ?thanks

Rainer_EMC · Accepted Answer

yes, that is safe to saywhether a checkpoint is mounted or not doesn't have anything to do with it working properly or being consistentThey became unmounted because at one point in time your savvol was running out of space and the system had to inactivate them.When a checkpoint gets inactivated its no longer consistent so it needs to get unmounted.When you refresh it, it does because usable again - like if you would delete and recreate it.The loophole there is that the scheduler or the refresh dont check if the checkpoint needs remounting.Take at look at the SnapSure manual for the exact rules as to what gets tried when a savvol runs out of space. If you have a lot of the you might have to adjust that global param about how much of the total space can be used by savvolsIts also a good idea to configure events for SVFS and VRPL - that could have alerted you in the first place when the savvol usage exceeds 90%

Rainer_EMC · Answer

an inactive checkpoint is dead - the only things you can do with it it are delete and refresh

yes, currently when a checkpoint gets auto-deleted that checkpoint file system does (has to) get unmounted and it doesnt get mounted automatically again when its refreshed by the scheduler.

This is planned for a future release. For now just mount it manually after it has been refreshed.

dynamox · Answer

yes, currently when a checkpoint gets auto-deleted
that checkpoint file system does (has to) get
unmounted and it doesn't get mounted automatically
again when its refreshed by the scheduler.

This is planned for a future release. For now just
mount it manually after it has been refreshed.

Rainer ...can you elaborate a little bit. What do you mean by auto-deleted ?

Thanks

dynamox · Answer

ok, that makes sense ...but why my 'inuse' column contains checkpoints that have 'n' next to them ? Do i interpret this correctly that it means checkpoint not in use ? What does it mean ?

Rainer_EMC · Answer

well, some people think checkpoints are static thing like CDROMs since they are readonly.
they arent - since they are just the changes to the active file system they are a "living" thing
When you change a block in the active file system that hasnt been saved in any checkpoint before the old value has to be tracked in the savvol

Thats why, if we run out of space on the savvol, we first try to extend the savvol.
If that doesnt work or if a global percentage used is reached the system will inactivate the oldest checkpoint in order to keep the others alive.

Thats what I've sloppily meant with auto-delete

Rainer_EMC · Answer

Oh, and it case it wasnt clear - a checkpoint works perfectly well if it isnt mounted but is active
You just cant access its data then through .cpkt or the Windows Previous Versions (SCSF).

You can still restore it or mount it again

Rainer_EMC · Answer

In this case inuse just means whether its mounted or not

dynamox · Answer

once that checkpoint gets refreshed, will it become mounted again ?Thanks

Rainer_EMC · Answer

once that checkpoint gets refreshed, will it become mounted again ?

no, currently it doesnt get mounted automatically again when its refreshed by the scheduler.

This is planned for a future release.

For now just mount it manually after it has been refreshed.

Or delete it - then I think the scheduler will have to recreate it and and it does get mounted again

dynamox · Answer

Or delete it - then I think the scheduler willhave to recreate it and and it does get mounted again yeah, this is the only file system i had running out of saVvol space ..all other file system checkpoints have 'inuse' is yes. I am curious to see if these checkpoints will be remounted once they are refreshed, i don't have anybody requesting restores at the moment so i will leave them unmounted. I have 14 days retention period for snapshots ..so they will be refreshed soon.

dynamox · Answer

well ..checkpoint got refreshed last night yet the 'inuse' status is still no, i went ahead and manually mounted one of the checkpoints, the question is ..will it remain mounted when it gets refreshed next time ?

dynamox · Answer

is it safe to assume that all existing checkpoints that have 'inuse' = no ...are ok and just need to manually remounted. I am just concerned about data integrity of those checkpoints ..since they became unmounted (i still don't understand why )

Rainer_EMC · Answer

yes it willthe scheduler will just issue a checkpoint refresh, which doesnt cause it to be mounted or unmounted

dynamox · Answer

They became unmounted because at one point in time
your savvol was running out of space and the system
had to inactivate them.
When a checkpoint gets inactivated its no longer
consistent so it needs to get unmounted.
When you refresh it, it does because usable again -
like if you would delete and recreate it.
The loophole there is that the scheduler or the
refresh dont check if the checkpoint needs
remounting.

ok, i see ..makes sense. I will go ahead and re-mount all the checkpoints.

Thank you for your help.

Rainer_EMC · Answer

I bet if your data mover log is long enough you'll find a message like this:

Sun May 13 06:06:55 CEST 2007
Error
SVFS
Slot 3: 1179029209: FSID:314 SavVol:588 Inactive

and before that the actions the Celerra tried to get more space

Celerra

Was this post helpful?