From: Michael Kershaw (mike.kershaw_at_gmail.com)
Date: Thu Oct 12 2006 - 01:37:15 CEST
Message-ID: <a7b43ddc0610111637i621c2ae3pc54e673a06a024f0@mail.gmail.com> Date: Wed, 11 Oct 2006 19:37:15 -0400 From: "Michael Kershaw" <mike.kershaw@gmail.com> Subject: Re: [suse-sles-e] Multipath blacklisting
Alexei,
Without blacklisting the lun, the device-mapper would still
retain possession of it. That being the case here, the HDS array
cannot remove the lun, due to lun protection built into the array. It
pretty much considers the device to be in use, and prevents any
potential disasters. With the appropriate uid's in place for the
blacklisting, i was able to flush them from the device-mapper, and
prevent them from being obtained. After those steps were taken, the
HDS array was able to remove them, and life went on! =] The
environment this is taking place in, is IBM xSeries servers,
qla2342's, Sles9 SP3, with fiber storage coming from various models
of HDS arrays.
Mike Kershaw
On 10/5/06, Alexei_Roudnev <Alexei_Roudnev@exigengroup.com> wrote:
> Why do you need to blacklist these LUN's instead of just evacuating data and
> deleting on the SAN system (or better from the LUN map on SUN system)?
>
> What happen if you remove LUN without blacklisting?
>
> ----- Original Message -----
> From: "Michael Kershaw" <mike.kershaw@gmail.com>
> To: <suse-sles-e@suse.com>
> Sent: Thursday, October 05, 2006 9:04 AM
> Subject: [suse-sles-e] Multipath blacklisting
>
>
> > Hi All,
> >
> > I've been using multipathing a solid 6 months now in a couple of
> > (18) node production 10g clusters. What I'm tasked with right now, is
> > deporting about 30 luns or so from one of the clusters for recycling,
> > and before we remove the luns from the SAN side of things, I want to
> > blacklist them from the device-mapper, and then remove them from the
> > OS. Problem is now, I can't get the blacklisting to work for me.
> > Have any of you out there been able to blacklist fiber luns via the
> > UID of the lun? As it stands, I'm trying with a configuration that
> > looks like this:
> >
> > devnode_blacklist {
> > wwid 3600096a1303030343030303030383262
> > devnode "^(dm-|fd|loop|md|ram|raw|scd|sr|st|sda) [0-9]*"
> > devnode "^hd[a-z] [0-9]*"
> > devnode sda # internal disk
> > }
> >
> > Being that I've never done this before, Is it safe to assume that if
> > in fact it was working, that the UID(or wwid) of that lun, would no
> > longer be present in the /dev/mapper/ directory? If i'm overthinking
> > things here, or just plain old missing something, I'd appreciate a
> > little guidance.
> >
> > Thanks In Advance,
> > Mike Kershaw
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: suse-sles-e-unsubscribe@suse.com
> > For additional commands, e-mail: suse-sles-e-help@suse.com
> >
> >
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: suse-sles-e-unsubscribe@suse.com
For additional commands, e-mail: suse-sles-e-help@suse.com
This archive was generated by hypermail 2.1.7 : Thu Oct 12 2006 - 01:37:22 CEST