I ran across this same problem when i shared the Nightly.0 for backup purposes. When the nightly.0 rolls over to the Nightly.1, the CIFS share keeps track and actually keeps the share mapped to nightly.1. (same thing next night when it rolls over to nightly.3). I posted the problem a few weeks ago and the only "solution" anyone could come up with was running an rsh command to stop and recreate the share every time the backup runs. I opted to forget the snapshot directory backup and go with a live backup of the C$\ share.

--Patrick Day

owner-toasters@mathworks.com on 03/01/99 01:34:39 PM Please respond to "DDA.RFC-822=tmerrill@mathworks.com/P=Internet/A= /C=us" @ X400 To: "DDA.RFC-822=toasters@mathworks.com/P=Internet/A= /C=us"@X400 cc: "DDA.RFC-822=tmerrill@mathworks.com/P=Internet/A= /C=us"@X400 Subject: nightly.1 busy: snapshot rollovers not happening

Folks, What does this "busy" mean?

bat> snap list Volume vol0 working....................................................

%/used %/total date name ---------- ---------- ------------ -------- 0% ( 0%) 0% ( 0%) Mar 01 11:01 hourly.0 0% ( 0%) 0% ( 0%) Mar 01 10:01 hourly.1 2% ( 2%) 1% ( 1%) Mar 01 09:01 hourly.2 14% (13%) 11% ( 9%) Feb 27 00:01 nightly.0 22% (15%) 19% (12%) Feb 26 00:00 nightly.1 (busy)

The nightly snapshot apparently didn't roll over, and hence a lot of testing suites that we run that depend on an image at mignight are giving us out of date results.

The word "busy" is not mentioned anywhere in the ontap html docs (NetApp Release 5.2.1: Thu Dec 31 12:56:45 PST 1998). I've searched the archives for this list, and hit upon a similar thread just a few weeks ago (RE: Shared Snapshot directory), but the discussion went off on backups, not why the rollover wasn't happening.

I suspect it's a CIFS locking issue. So, is there a way to *force* the automatic nightly snapshot to rollover? If not, I guess I'll just hand-job this to do "manual automatic" snapshots at midnight and `snap rename` the busy ones. <sigh> What's the use of having the automated capability if it doesn't work when files are locked, he sarcastically muttered under his breath... ;)

And, if nightly.1 is busy, how come nightly.0 wasn't just replaced with one dated "Mar 01 00:01" instead?

I realize the implications to possible Windows clients: programs dying, BSODs, etc., but that's okay in this environment. And, since these areas are read-only, there should be no data loss, yes?

And, how TF do I get rid of this without rebooting the filer?

bat> snap delete nightly.1 Snapshot in use by dump or a CIFS share.

It's ain't being dumped, and never has been, so it's a CIFS share issue.

(`cifs sessions *` works. Wo-hoo! But, which one is holding that particular snapshot busy?)

(1-month newbie to NetApp filers, still learning the intricacies... Thanks for listening.)

Until next time...

Todd C. Merrill The Mathworks, Inc. 508-647-7792 24 Prime Park Way, Natick, MA 01760-1500 508-647-7012 FAX tmerrill@mathworks.com http://www.mathworks.com ---