JB's Miscellaneous Ramblings

LSE's current hardware is dying. LSE's new hardware will HOPEFULLY be ready soon, but it isn't yet.

If you're trying to contact anyone on LSE, you might want to use an alternative address for a while.

For me, gmail (jailbait) is a good bet.

{sigh}

Thoughts of 'where are they now' are coming to mind... might as well ask the net at large.

I'd have more than the one name right now, but it really seems like LSE's disk is failing.
FRAK.

Anyone trying to get in touch with any of our users in the coming days might want to use alternative addresses...
(For me, jailbait/gmail)

In the mean time, anyone ever heard of Pam Israel from Poughkeepsie?

(The following post is edited together from an afternoon's worth of IRC conversation on 2 different geekly channels. I'm sorry if I failed to make it clear, but I've been staring at all of this too long to have any distance left.)

Is there anyone here with netapp experience, especially as concerns space allocation and snapshots, especially as concerns iSCSI? I've been trying to get meaning out of the docs for a while now and I just keep glazing over... and not in the good Krispy Kreme glaze sort of way.

I'm trying to figure out how much diskspace I need to reserve where and via which mechanism to have my snapshots A) snap and B) be usable in a lun clone as necessary for data recovery.... and to figure out if I'm even thinking about all of this in the right way. NetApp's docs... well, they don't totally blow, but they're very circularly referential. If there's a proper path through dealing with their concepts, I've yet to find it.

I've got a filer with an aggregate with a flexvol with a lun.
I need to figure out appropriate allocation of reserve space to let me make snapshots and then to let me use them, if needed.
Despite that, on a 350GB volume, the lun in question is 200GB and has only 120G used (at a reasonable ROC), I'm only managing to be able to make 1 snapshot.

I've got 10% snapshot reserve on the volume - 35G.

carbarn*> snap delta mysql
. . .
From Snapshot To KB changed Time Rate (KB/hour)
--------------- -------------------- ----------- ------------ ---------------
nightly.1 Active File System 282316 1d 16:47 6922.337

I'm really just not getting how much I need to be reserving here... and where to reserve it. 70mb/hr =560mb/8hr, 1.6GB/day., 840mb/12h. So to keep 6 snaps from 8hr intervals and 2 dailys, I'd expect to need something lik 8GB. That's not very much.

What is using up the space between my 200GB lun and my 350/10% volume?
(There's nothing on the vol but the lun and the .snapshot directory/contents.)

Alternately, how do I tell how much space I need to have available to make a snapshot Right Now?

carbarn*> df -rh mysql
Filesystem total used avail reserved Mounted on
/vol/mysql/ 315GB 315GB 0KB 140GB /vol/mysql/
/vol/mysql/.snapshot 35GB 10GB 24GB 0KB /vol/mysql/.snapshot

... and yet, I've got only one nightly snap.

As well, among many other things, I'm unclear what the 140GB reserve on the volume is for. The only thing on the volume is one 200GB lun. Given that the LUN is space-reserved, I'd EXPECT to see 200GB reserved and 115GB free. I have no idea what the 140GB reserve is all about.

I'm developing an unfortunate feeling that, although I understand LUN snapshots to be block-level (and I have 3rd party confirmation to that effect), they require the full size of the lun in available diskspace. This is... bad... if it's true.

Anyone? This is also going in to a support request with NetApp, but given how my last one of those went, I'm not amazingly hopeful. (The local folks are good. The Indian outsourced tech-support crew, not so much... and it may just be language issues, but I can't really tell.)

(The filer's a 270c running DataONTAP 7.1 and the heads are named "Wellington" and "Carbarn", as our naming scheme is T stops and using the orange line maintenance yard for the storage host seemed to make sense to me. :)