Unwanted increase of LUN size

[LKULCSAR2]

Hi There,

I have a theoretical question. Let's imagine that There is a LUN on a volume without any snapshot. The LUN is utilizing 95% of the usable volume space. Do you know any any circumtances when the size of a LUN can increase or go offline because of a filesystem full on the connecting host (Windows 2003/2008, RedHat Linux, AIX servers)?

As I know it's not possible, but who knows:-)

Thank you for your responses in advance!

Laszlo

[JGPSHNTAP]

If the LUN is thick provisioned and there are no snaps on the volume with fractional reserve at 0, your Lun will never go offline.  Now, the LUN might run out of space on the OS side, but that's another story

[aborzenkov]

It's possible. NetApp needs extra space for new (over)writes before it reclaims old blocks. So with very intensive overwrite activity (overwrite from NetApp perspective) it is possible that it won't have time to free space. Highly unlikely in real life, but could be triggered in lab conditions.

[GUTLESBACHER]

I'm pretty sure the only thing that would be affected under heavy load/overwrite activity would be performance. Not enough space for certain operations may slow down the operations, but I know of none that would make a LUN go offline as long as snapshots (volume based not snaps on the LUN) are not involved.

Overall for the initial question, a LUN or volume will never go offline just because the filesystem of the host is full as long as the LUN / volume on the NetApp is not full. This might cause the OS to get timeouts, errors or take it offline from the OS side, but on NetApp side if no other space in the volume is used (for example because of snaps) then a thick LUN thats 95% the size of a volume will never exceed its allotted space therefore the volume can never get "full"

Of course a full LUN or volume often corresponds to the host filesystem getting full, but you have to see them as two seperate issues.