On Fri, May 04, 2007 at 08:55:54AM -0700, Mike Kupfer wrote:
> >>>>> "stevel" == Stephen Lau <[EMAIL PROTECTED]> writes:
> 
> stevel> It looks like there was an old errant hg process holding onto
> stevel> onnv/onnv-gate, so it never got unmounted (and thus remounted).
> 
> stevel> I've killed it now, which forced the remount, and it looks like
> stevel> it should be serving the correct onnv-gate now.
> 
> There's more than one onnv-gate?  I don't understand how holding the
> filesystem mounted would cause problems.

When I have to do a rollback, I clone the old onnv-gate into the "new"
onnv-gate up to the common ancestor changeset.  I then push the new
rebuilt changesets on top of that, move the old onnv-gate aside, and
drop the new onnv-gate onto the "/hg/onnv/onnv-gate" name.

Unfortunately, anon tends to have enough traffic of people cloning it
that the loopback mount of onnv-gate into /home/anon/hg/onnv/onnv-gate
doesn't always timeout, so it has the mount pointing to the old
workspace instead.

It's usually a simple matter of killing whatever is using that mount
(and thus zorking some poor sap's current bringover, but given that it's
a bad bringover anyway, i don't consider that too bad), umounting the
/home/anon/hg/onnv/onnv-gate, and allowing it to be remounted by the
next anon clone.

cheers,
steve
-- 
stephen lau // [EMAIL PROTECTED] | 650.786.0845 | http://whacked.net
opensolaris // solaris kernel development
_______________________________________________
tools-discuss mailing list
[email protected]

Reply via email to