[CentOS] CentOS and Dell MD3200i / MD3220i iSCSI w/ multipath

Sat Jan 22 09:18:50 UTC 2011
Dr. Ed Morbius <dredmorbius at gmail.com>

on 20:07 Fri 21 Jan, Ross Walker (rswwalker at gmail.com) wrote:
> On Jan 21, 2011, at 7:20 PM, Edward Morbius <dredmorbius at gmail.com> wrote:
> 
> > 
> > 
> > On Fri, Jan 21, 2011 at 3:58 PM, Ross Walker <rswwalker at gmail.com> wrote:
> > On Jan 21, 2011, at 6:41 PM, Edward Morbius <dredmorbius at gmail.com> wrote:
> > 
> >> We've been wrestling with this for ... rather longer than I'd care to admit.
> >> 
> >> Host / initiator systems are a number of real and virtualized CentOS 5.5 boxes.  Storage arrays / targets are Dell MD3220i storage arrays.
> >> 
> >> ...
> > 
> > 
> > Once this is installed you need to setup dm-multipath, look for multipathd.conf in /etc, get the product id and vendor id from dmesg after making an initial connection via open-iscsi and use that in the mutipath config. Your going to need to use path utility 'rdac' in the config instead of tur.
> > 
> > Google is your friend here.
> > 
> > -Ross
> > 
> > 
> > _______________________________________________
> > CentOS mailing list
> > CentOS at centos.or
> > 
> > /etc/multipath.conf appears to be appropriately configured (we'd installed the MDSM host components):
> > 
> > ------------------------------------------------------------------------
> >          device {
> >                 vendor                  "DELL"
> >                 product                 "MD32xxi"

<...>

> AFAIK the RDAC you have installed looks correct and the config also looks
> good.

Thanks.

> Did you start the multipath service make a connection to each IP and do a
> 'multipath -ll' and see what shows up?
 
Yes, and yes.

We've actually run some fairly intensive disk tests (bonnie++ and a few
tens of thousands of 100MB file copies of random data) with no errors
across various hosts.

The on-connect errors are the biggest issue we've got, though general
concensus seems to be that we can ignore these.

What's moderately maddening is the lack of any clear documentation or
guidance, from Dell, RH, or the upstream open-iscsi / multipath
projects, on what we should be experiencing, and what, if any, errors
are considered "normal".

Think we've got a handle on it, but we're checking our sanity as well.

-- 
Dr. Ed Morbius
Chief Scientist
Krell Power Systems Unlimited