write to lun failed due to lack of space error

[csboradba]

Hi,

I am periodically having an issue where a LUN is taking itself offline with a "write to lun failed due to lack of space error". The windows server that has access to the LUN reports that there is 4 GB of free space available. In addition our monitoring tool reports 4GB of free space at the point the LUN goes offline. We do not run snapshots on the Volume however another volume in the same aggregate does. I have laid out the configuration at the end of the post.

The offline error has occurred 3 times over the last month although I can't see a particular pattern to the failure. My knowledge of NAS in general is very limited but from trawling through the configuration I have noticed that "space reserved" is turned off on the LUN. The online document states "It is recommended that you create a LUN with space reservations enabled so that blocks in the LUN can be updated when they are written to". Could this be the cause of the LUN running out of space and going offline? If my calculations are correct the volume is 12.7 TB and the LUN is around 12.69 TB so the LUN shouldn't be running out of space?

Has anyone any idea from my configuration what could be the problem or, if it is the space reserved setting why it is running out of space when the volume is big enough to hold the LUN?

Any help would be appreciated.

LUN
Description	NL LUN
Size	13635859545
Space Reserved	OFF

Volume
Name	NL
Type	Flexible
Status	Online,raid0
Used Capacity	12.7 TB
% Used	100%
Total Capacity	12.7 TB
Number of files	107
Max files	31.9 m
SNAP	NO
Fractional reserve	100%
Containing Aggregate	AGR1
Space Guarantee	Volume
Total Size	12.7 TB
Max directory size	20.5 MB

Aggregate
Name	AGR1
Type	Aggregate
Status	Online,raid0
Used Capacity	100%
Total Capacity	13.3 TB
Number of files	107
Max files	31.1k
Raid Size	9
Checksums	Block
Number of disks	42

There is also an additional volume on the aggregate.

Volume
Name	PD
Type	Flexible
Status	Online, raid0
Used Capacity	115GB
% Used	19%
Total Capacity	600 GB
Number of files	138k
Max files	20.8m
SNAP	YES
Fractional Reserve	100
Space Guarantee	Volume
Total Size	600 GB
Max directory size	20.5 MB

[ostiguy]

If the LUN is thin provisioned, what the host sees is irrelevant - the host has been allocated capacity it has not consumed yet. What happens when it attempts to consume it?

Well, it depends on how the lun and flexvol are configured - if the lun is not space reserved, space will be taken from the flexvol which it sits in. It looks like the flexvol is full - this is why the writes are failing. Can you shrink the other flexvol, and grow this one?

[csboradba]

Thanks for the quick reply. Yes, I will shrink the other flexvol and growth the thin provisioned one. I'm going to turn space reserved on too.

[mkopenski]

It looks like the aggregate is full

What does the output of 'df -hA' show?

If there are snaps on the aggregate you can turn those off and reclaim some of the reserve space on the aggregate.

[csboradba]

Yes, it defintiely is. I have investigated a bit further and the lun was over commited by 174MB. Since the aggregate was full this caused the lun to go offline. Fortunately I can shrink down another lun in the agregate so will use that space to increase the over commited lun and turn on space reservation.

Thanks for all your help.