multipath.conf for DELL MD3200 storage array

Unfortunatly I could not find a current DELL documention how to build a valid multipath.conf when using a MD3200 storage array. The config coming with the DELL software reports some errors:

Aug 09 14:22:19 | multipath.conf +104, invalid keyword: polling_interval
Aug 09 14:22:19 | multipath.conf +112, invalid keyword: prio_callout

/etc/multipath.conf


device {
vendor “DELL”
product “MD32xx”
path_grouping_policy group_by_prio
prio rdac
polling_interval 5
path_checker rdac
path_selector “round-robin 0”
hardware_handler “1 rdac”
failback immediate
features “2 pg_init_retries 50”
no_path_retry 30
rr_min_io 100
prio_callout “/sbin/mpath_prio_rdac /dev/%n”
}

The problem with the polling_interval is that it should be in the defaults section and not in the device section. But I don’t know what to do with the prio_callout. As far as I know the parameter is deprecated but the parameter is still documented in the “Storage Adminstration Guide” for SLES11 SP2. Is the documentation out-dated? How to configure it right?

Thanks!

[QUOTE=A13923;6134]Unfortunatly I could not find a current DELL documention how to build a valid multipath.conf when using a MD3200 storage array. The config coming with the DELL software reports some errors:

Aug 09 14:22:19 | multipath.conf +104, invalid keyword: polling_interval
Aug 09 14:22:19 | multipath.conf +112, invalid keyword: prio_callout

The problem with the polling_interval is that it should be in the defaults section and not in the device section. But I don’t know what to do with the prio_callout. As far as I know the parameter is deprecated but the parameter is still documented in the “Storage Adminstration Guide” for SLES11 SP2. Is the documentation out-dated? How to configure it right?
[/QUOTE]

I suspect there is some small error somewhere in your multipath.conf (due to a section not being closed correctly with the “}” for example).
It would be better to post the complete multipath.conf contents so we can see what’s happening as a whole.

Have you installed all the packages (also ones from Dell where applicable) as per Dell documentation?

Dell’s configuration guide I just Google’d (outdated though) shows this sample configuration for SLES 11 (the GM release) - see the layout quoted below.

If you append this to the multipath.conf sampels you can find on the server in /usr/share/doc/packages/multipath-tools, you should be good. Look for setting all the correct dots and comma’s - so to say. Otherwise you can get these types of configuration errors outputted as multipathd tries to execute settings in wrong sections.

& also see : http://doc.opensuse.org/products/draft/SLES/SLES-storage_sd_draft/multipathing.html

Quote from the Dell doc I mentioned:

Cheers,
Willem

Hello Willem,

the multipath.conf is the same. As I mentioned it seems that “prio_callout” is deprecated

http://christophe.varoqui.free.fr/

“prio = rdac” is already set but do I need “prio_args” too? I couldn’t find a example with rdac and “prio_args”.

Back to “prio_callout” the mentioned “/sbin/mpath_prio_rdac” isn’t existing on my system. Do I have to install additional software?

I have checked the brackets this shouldn’t be the problem.

Thanks
Achim

EDIT: According to the following link to the redhat knowledgebase the parameter is now “just “prio””. So could I delete “prio_callout” and everything is fine?

http://www.redhat.com/archives/dm-devel/2012-March/msg00155.html

You are correct as this file is not included (as far as I can find) in the init/mpath packages any more.

[QUOTE=A13923;6150]EDIT: According to the following link to the redhat knowledgebase the parameter is now “just “prio””. So could I delete “prio_callout” and everything is fine?

http://www.redhat.com/archives/dm-devel/2012-March/msg00155.html[/QUOTE]

Yes, that does make sense (also reading Bryn’s reply), so I’d leave it out.

If your paths are getting the correct id’s and are equal across connected systems (if clustering is applicable), you should be fine.

To be sure so some testing with at least one test partition and open content on it to see what happens when pulling one of the connections to the storage and then switching them (to make sure path failure and return gets handled correctly). If it all works as expected, you should be good AFAIK.

I do agree it’s confusing, as the Red Hat documentation does state clear changes in version 6, but the SLES documentation (that I’ve found) does not state anything about this but still generally refers to it as working : http://www.suse.com/documentation/sles11/singlehtml/stor_admin/stor_admin.html

I’ll drop this question to my Novell contact.

Cheers,
Willem