When lun space is consumed in the lun the first time, the data in the volume will grow until the 900 Gb of the lun space has been used. When you free up room in the lun, the WAFL filesystem can't know this, so all those 'freed blocks' will be kept in use on WAFL. When however it is a windows filesystem and you use Snapdrive 5.0, you can enable space reclamation. Snapdrive will then scan your NTFS filesystem and tell WAFL which blocks can be freed. Then and only then free space in the lun will be free space in the volume.

What Andrey says is correct. When you only have one snapshot in the volume, it is possible that all data will have changed since the snapshot was taken so the snapshot will also be 900Gb. This is idd worst case scenario.

What I would try is to just disable snapshot reserve (set it to 0%). This is the best practice when using luns. Snapshots will still be taken, but will just use free space in the volume.

Grtz,
Tom
Uptime Belgium


-----Original Message-----
From: owner-toasters@mathworks.com [
mailto:owner-toasters@mathworks.com] On Behalf Of Oliver Bassett
Sent: donderdag 3 januari 2008 0:16
To: toasters@mathworks.com
Subject: RE: Incredible shrinking volume

Can't remember quite when this changed (7.0?) but it now only reserves 100% of the used space within the LUN when the first snap is taken. However as data changes the space used by the snapshot will grow, and if the LUN is slowly filling then the reserve space will slowly be going up also.

Thanks
Oliver Bassett


-----Original Message-----
From: owner-toasters@mathworks.com [
mailto:owner-toasters@mathworks.com] On Behalf Of Glenn Dekhayser
Sent: Thursday, 3 January 2008 10:50 a.m.
To: Borzenkov, Andrey; Stephen C. Losen; toasters@mathworks.com
Subject: RE: Incredible shrinking volume

Hang on there.  If you have a space-reserved LUN, then as soon as snapshot #1 is taken, Ontap immediately takes 100% of the LUN space at that time.  Additional snaps do not matter to LUN space reservation.

-----Original Message-----
From: owner-toasters@mathworks.com [
mailto:owner-toasters@mathworks.com] On Behalf Of Borzenkov, Andrey
Sent: Wednesday, January 02, 2008 3:57 PM
To: Stephen C. Losen; toasters@mathworks.com
Subject: RE: Incredible shrinking volume

Check with "df -r" for reserved space. I am not sure whether synchronous SM is based on snapshots (description is a bit vague), but if yes - by default creating snapshot on volume with LUN will reserve the amount of space equal to changed data. The more data is changed in LUN, the more reserved space you need. In the worst case you need 1.8TB volume for your 900GB LUN.

§³ §å§Ó§Ñ§Ø§Ö§ß§Ú§Ö§Þ / With best regards / Mit freundlichen Gr¨¹¦Âen

---
Andrey Borzenkov
Senior system engineer

-----Original Message-----
From: owner-toasters@mathworks.com [
mailto:owner-toasters@mathworks.com] On Behalf Of Stephen C. Losen
Sent: Wednesday, January 02, 2008 9:53 PM
To: toasters@mathworks.com
Subject: Incredible shrinking volume


I have a fas270 with a volume named "san" containing one qtree and
one LUN within the qtree.

There is only one aggregate and it contains volumes "san" and "vol0"
(root).

The san volume is being synchronously snapmirrored to another
fas270.  Here are the sizes of everything:

df -A

Aggregate               kbytes       used      avail capacity
aggr0               1253187072 1200347772   52839300      96%
aggr0/.snapshot              0          0          0     ---%

df

Filesystem              kbytes       used      avail capacity  Mounted on
/vol/vol0/            16777216     349732   16427484       2%  /vol/vol0/
/vol/vol0/.snapshot    4194304      64892    4129412       2%  /vol/vol0/.snapshot

/vol/san/           1119669456 1048001824   71667632      94%  /vol/san/
/vol/san/.snapshot    58929968     102176   58827792       0%  /vol/san/.snapshot

lun show

/vol/san/vmail/lun0        900.1g (966503301120)  (r/w, online, mapped)


I got an autosupport email saying that the san volume had run out of space,
and it was indeed 100% full.

I grew the san volume a little bit and I also reduced the snap reserve
to 5%.

Now as I watch the "df" output the san volume continues to slowly
consume space.  As I understand it, the LUN is a fixed length file,
so it cannot be growing.  I can only conclude that WAFL metadata files
in the volume must be growing, perhaps as a result of the LUN being
populated with data, or the snapmirror, or both.

I'm concerned that I'll run out of space again in the volume, at which
point I am just about out of options for enlarging it.

Does anyone know what's going on here?

As I have written this email, the space available in "san" has
gone from 71667632 down to 71245556 and it's been dropping
slowly and steadily for hours.

Does anyone know what's going on here?  Will I hit bottom before
running out of space again?

Steve Losen   scl@virginia.edu    phone: 434-924-0640

University of Virginia               ITC Unix Support





The information contained in this email is privileged and confidential and
intended for the addressee only. If you are not the intended recipient, you
are asked to respect that confidentiality and not disclose, copy or make use
of its contents. If received in error you are asked to destroy this email
and contact the sender immediately. Your assistance is appreciated.