Summary: LU and Zones issue

Sorry the summary is so long in coming...

>        I am in a bit of a bind. I have a system with 5 NG Zones and
> need to perform a Live Upgrade. The first lucreate failed (and I think
> I've addressed the problem), but I have some leftover cruft which is
> preventing another lucreate from succeeding. I cleaned up as much as I
> could, but I have an errant zsched process that is tied to an LU Zone.
> I can't kill that zsched (even -9) and it is preventing me from
> removing the last loopback mount created during the failed lucreate.
> Until I remove that last loopback mount I don't think another lucreate
> will work.
>
>        So, has anyone ever had to clean up after a failed lucreate
> with NG Zones and figured out how to do this ? I can't reboot (that is
> why I need to be this as a Live Upgrade). Please respond directly to
> me and I'll summarize for the list.

The short answer is that I had to reboot to clear the issue. The crux
of the problem was not the hanging mount, but the hanging zsched and
phantom running NG Zone that was keeping the FS from being unmounted.
I received the following suggestions (and my responses). I am
including these for completeness and if they can help anyone else.

I will also try these steps as I have run into the same problem again.
If I learn anything new this time I will post a revised summary.

Thanks for all the suggestions, they are very welcome.

On Thu, Oct 21, 2010 at 8:42 AM, Rob De Langhe
<rob.de.langhe@twistfare.be> wrote:

> what if you run "ptree" against that "zsched" process? Has it any
> subprocesses running on which it is waiting?

No parents or children. It is the zone scheduler for the virtual zone
booted by the LU process to work on the NG Zone. Unfortunately,
`zoneadm list` does not know about it or I would just halt it.

On Thu, Oct 21, 2010 at 11:03 AM, Rob De Langhe
<rob.de.langhe@twistfare.be> wrote:

> no effect of "kill -9" sounds like it's hanging in some driver.
>
> Did you unmount manually some of the loopback mounts while the "lucreate"
> was still running ?

Nope, long after the lucreate had died a horrible death.

On Fri, Oct 22, 2010 at 4:27 AM, Rob De Langhe
<rob.de.langhe@twistfare.be> wrote:
> can you run a "df -k" and a "mount -v", to see what filesystems are
> (still) mounted from the LU ?

Never got to this as we had already rebooted.

On Fri, Oct 22, 2010 at 9:22 AM, Sergey Prilutsky
<sprilutsky@hotmail.com> wrote:
> Several things here:
>
> 1. When you run lucreate, always let it finish running, even if you see
> errors problems, then use ludelete to remove new BE
>
> 2. To fix you problem now, try this:
>
>  # mv /etc/lutab /etc/lutab.old
>
> # cd /etc/lu
>
> # mv ICF.1 old_ICF.1
>
> # mv ICF.2 old_ICF.2

Never got to this either as the system had been rebooted.

--
{--------1---------2---------3---------4---------5---------6---------7-------
--}
Paul Kraus
-> Senior Systems Architect, Garnet River ( http://www.garnetriver.com/ )
-> Sound Coordinator, Schenectady Light Opera Company (
http://www.sloctheater.org/ )
-> Technical Advisor, RPI Players
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers