Start a Conversation

This post is more than 5 years old

Solved!

Go to Solution

1442

November 10th, 2014 15:00

Release Celerra File System Capacity

Is there either a scheduled task or a manual method for "releasing" capacity within a Celerra file system? In this case, an iSCSI attached LUN filled up which in turn caused the file system to fill up. A bunch of stuff was deleted in the LUN, but the file system is still full. I haven't been able to find any documentation that speaks to this specifically. Let me know if you need any other details. Thanks

1 Rookie

 • 

20.4K Posts

November 11th, 2014 06:00

you can't free up that capacity, the iSCSI LUN has "balooned" to take up the free space in the file system. Just because you deleted data in the iSCSI LUN, it will not automatically shrink the iSCSI LUN (thus freeing up space in the file system). There is no such thing as VAAI thin block reclaim/UNMAP on Celerra that would release any unused blocks back to the file system.

When I used to deploy iSCSI LUNs on Celerra, i always had a dedicated file system for each LUN so i am curious why are you concerned about freeing up that space ? Even if you were able to get that space released, you still would not be able to release it back to the storage pool as you can't shrink file systems.

1 Rookie

 • 

20.4K Posts

November 10th, 2014 16:00

there is no way to shrink an iSCSI LUN that i am aware of.

8.6K Posts

November 10th, 2014 17:00

Not within the LUN itself (for Celerra ISCSI which is pretty old)

Check if there are snapshots on the LUN

If its vmWare you could storage vMotion to a new LUN

1 Rookie

 • 

20.4K Posts

November 11th, 2014 06:00

no, iSCSI LUN is a file inside of a file system, it will not shrink.

86 Posts

November 11th, 2014 06:00

Thanks, dynamox. That makes sense. I'm not sure if this particular file system only contains a single LUN or not. I'll find out. What I'm picturing in my head is something similar to a block LUN within a storage pool. The used data within the storage pool is only equal to the data being written within the underlying LUNs.

86 Posts

November 11th, 2014 06:00

No, not asking about decreasing the size of the LUN. The file system is still full. If data has been deleted in the underlying LUN, should the parent file system also gain additional free capacity?

1 Rookie

 • 

20.4K Posts

November 11th, 2014 06:00

Clayton wrote:

What I'm picturing in my head is something similar to a block LUN within a storage pool. The used data within the storage pool is only equal to the data being written within the underlying LUNs.

modern arrays using VAAI unmap sure, not on an ancient Celerra

86 Posts

November 11th, 2014 06:00

Let me rephrase because I don't think my question is clear. I'm not asking if I can shrink a file system or a LUN. I do not want to change the size of either. Let me use example numbers. Let's say I have a 100GB file system & a 100GB LUN underneath it. At one point the LUN filled to capacity and therefore the FS was also full. Data was freed up on the LUN so now its down to 50GB, but yet the file system is still full. My question is, how can I "release" or "free up" that capacity within the file system? My thought was that this should be done by some kind of automated cleanup job.

8.6K Posts

November 12th, 2014 04:00

Celerra ISCSI has been developed before VAAI unmap was available

86 Posts

November 17th, 2014 07:00

Is there any documentation that explains this in detail? It doesn't make sense that if you delete data within a LUN, that space would not be freed up within the file system. I would like to better understand how that works at the file system level.

8.6K Posts

November 17th, 2014 08:00

Another example – if you have Outlook with a PST file and delete an Email your PST file size doesn’t change – even if you empty the deleted items folder

Only if you reorganize it then the PST file will shrink

8.6K Posts

November 17th, 2014 08:00

Deleting data within a LUN isn’t really a delete to the underlying server file system.

All the clients file system does is change its metadata that these blocks are now free

Unless you have an implementation that also tells the underlying block or file layer to free that block (like via SCSI TRIM/UNMAP) – its not considered free there

No Events found!

Top