ONTAP Discussions

SIS START -s error help!

macmillangm
6,898 Views

I come from an environment that ran an EMC san, so my NetApp skills are about a noo-b. Wow, what a radical difference!

Anyway....

I have an agg1 with 2 volumes. One vol is 1024gb and the other is 200gb. I have a LUN on vol1 that is sized to the entire volume and in the second 200gb volume I have a lun that is 100gb with the remaining 100gb unused.

I enable sis and attempt to start when I get an error "Aggregate aggr1 is full Change logging metafile on SIS volumes in the aggregate can not hold any more fingerprint entries."

I have data on these LUNS where one is a VMFS data store and the other 1024gb LUN is an RDM to a VM file server.

6 REPLIES 6

pascalduk
6,776 Views

Deduplication has some overhead because it needs to store metadata. In your case it seems there is not enough space to store the metadata. The implementation guide for dedupe (http://media.netapp.com/documents/tr-3505.pdf) has the following guideline:

The guideline for the amount of extra space that should be left in the aggregate or volume for the deduplication metadata overhead is as follows:


• If you’re running Data ONTAP 7.2.X, leave about 3% extra space inside the volume on which you plan to run deduplication.
• If you’re running Data ONTAP 7.3, leave about 1% extra space inside the volume on which you plan to run deduplication, and around 2% extra space outside the volume in the aggregate.

macmillangm
6,776 Views

Ok, I have the LUN's set as reserved with about 700gb out of 1000gb used. If I remove the check for "reserved" I will potentially reclaim enough space to run SIS START but will I risk losing or damaging the 700gb of data in the LUN?

Fot the record, Im running 7.3

keitha
6,776 Views

George,

I ran into this not that long ago. It likely isn't the metadata that is preventing this but rather the snapshot reservations. Do you have any existing snapshots on these volumes? If you don't and the fractional reserve is set to 100% then when you start ASIS it will try to take a snapshot and reserve a additional amount of space for the LUN. If you set the reserve to 0 then you should be OK to start. You won't damage the data in the LUN as if you do run out of space ONTAP will offline the LUN before any damage occurs. That would take your VMs offline though. From what you described though it sounds like it is the reservations not the metadata filling up the aggr though.

Keith

macmillangm
6,776 Views

Ok, I went through the SNAP settings and basically cleaned house, since Im not snapping anyway.

Do I need to delete any pre-existing snapshots? I dont know if I have any and I dont know how to tell if I do.

So, I continue to see the same error, but I can change the "space reserved" setting on the LUNS.

When I do, and attempt to apply it I get a pop up "Reducing the size of a LUN may result in data corruption. Do you still want to continue?"

I am guessing this is a standard warning and I wont have any issues however this is production.

The Volume on this is 1024GB and the LUN reservation is 100%. The actual data on that lun is about 700GB.

So if I change the space reservation, in theory, I should not suffer any consequences or data loss.

Unless the snapshots somehow throw a wrench into the whole thing.

keitha
6,776 Views

It's not so much the space reservation on the LUN but the fractional reserve. If the fractional reserve is set to 100 then when you snapshot the volume ONTAP will reserve equal amounts of space for the the snapshot as the LUN size. In your case, if your LUN is 1023GB then it will try to reserve an additional 1024GB which I suspect you don't have on your aggergate. From the command line you can set this to 0 so that it will not reserve any space. However you will have a snapshot open while you dedupe so you should have suffcient space in the volume to hold the changes made to the volume while the controller is deduping. Do you have any free space on the aggeragte at all? If so, you should grow your volume so you have some space.

You might want to start with the smaller volume since it has space and that will likely free up some room on the aggr which might help you dedupe the larger volume. The first ASIS pass will take a while so you might want to plan that one for a weekend if your controller is busy. Things are a little tight on there!

Scott Lowe has a good blog about some of these settings here http://blog.scottlowe.org/2007/12/05/managing-lun-space-requirements-with-netapp-storage/

macmillangm
6,776 Views

Do you have any free space on the aggeragte at all? If so, you should grow your volume so you have some space.

No, Im up to my eye balls, aggerates have no space. If I had budget, I'd buy more disk. As it is, I cant get them to buy coffee

I can start with the smaller LUN's (volumes) no problem.

I removed the "spaced reserved" setting and the following error stopped filling the syslog.

[[NYFAS02: sis.aggregate.full:warning]|javascript:popEMS('sis.aggregate.full');]: Aggregate aggr1 is full Change logging metafile on SIS volumes in the aggregate can not hold any more fingerprint entries.

Tue Jan 20 16:47:20 EST last message repeated 53 times

However the Frac Reserve is still 100. I should still set that to zero?? vol options

<volname>

fractional_reserve 0

Fractional Reserve :

100

Public