[OpenIndiana-discuss] auto-scrub and its result

[Richard L. Hamilton]

On Dec 29, 2012, at 6:57 PM, Bob Friesenhahn wrote:

> On Sat, 29 Dec 2012, Jim Klimov wrote:
>> 
>> Well, apparently Bob does use the copypasted line and it works for him ;)
> 
> Yes, and it should.  There is no need to add more complexity than is warranted.  Even if the scrub takes a very long time and scrub is requested again while scrub is still running, the Solaris 10 documentation says that the request will return an error and the existing scrub will continue running.  The OpenIndiana documentation does not say what happens if scrub is requested while scrub is still running but it is easy to verify that OpenIndiana works the same:
> 
> % pfexec zpool scrub rpool
> % pfexec zpool scrub rpool
> cannot scrub rpool: currently scrubbing; use 'zpool scrub -s' to cancel current scrub
> 
> Bob

I seem to recall that only one scrub could be running at a time, even against different zpools.  So the cron that does them all might never actually do more than the first.

Another possible problem (maybe long since fixed): I had a tiny file-based zpool and tried to scrub that and another:

zpool scrub mpool;sleep 2;zpool scrub rpool

(where mpool is the one based on a file, itself on zfs) and that crashed my system.  Of course, said system is far back (snv_97), so maybe it's been fixed since then.  And for additional weirdness, I think mpool was originally created with zfs on a Mac (the file having been shared via iSCSI).  Yes, I do some weird things...but there was a reason at the time...

-- 
eMail:				mailto:rlhamil at smart.net
Home page:			http://www.smart.net/~rlhamil/
Facebook, MySpace,
AIM, Yahoo, etc:		ask