Need help with powerpath.

[Silver_1]

We have a server connected the EMC Clarrion and the SAN team taken away some LUNs and give back some LUNs.

How can i restart the powerpath in HP UX.

I searched /sbin/init.d, couldn't find anything related to powerpath ...

ANy help is appeciated ...

Tx,
Nair

[Geoff Wild]

Shouldn't have to restart (only way anyways is a reboot).

This might work:
symmask -sid XXXX refresh

where XXXX is the frame number.

You can do a:

symcfg discover

to find out the frame number

As I have said before - if the SAN team does their job, you shouldn't have to do anything...

Rgds...Geoff

[Silver_1]

My actual intension is to remove the dead paths. I have used the # powermt check but it says like this

Warning: CLARiiON device path unknown is currently dead.
Do you want to remove it (y/n/a/q)? y
Illegal device unknown.
Volume not found.


it shows the path as unknown.....

here is the powermt display dev=all output.

CLARiiON ID=222222
Logical device ID=3333333 [LUN 2004]
state=alive; policy=CLAROpt; priority=0; queued-IOs=0
Owner: default=Unknown, current=SP B
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
13 0/0/8/1/0.1.11.0.0.0.5 unknown SP A1 active dead 0 1
12 0/0/8/1/0.1.7.0.0.0.5 unknown SP B1 active dead 0 1
15 1/0/8/1/0.1.11.0.0.0.5 unknown SP A0 active dead 0 1
14 1/0/8/1/0.1.7.0.0.0.5 unknown SP B0 active dead 0 1

[Geoff Wild]

Try this:

powermt remove force dev=cXtXdX

Where cXtXdX is your dev you want ot remove...

If that fails, reboot may be the only option :(

Rgds...Geoff

[Stuart Abramson]

In general, you go through the following sequence anytime EMC disks are added, removed or changed:

powermt display paths
powermt restore
powermt check
powermt config
powermt save
powermt display paths

You should be able to look these up with "man powermt". If you can't, then check your MANPATH.

This will work every time with Symmetrix disks. CLARiiON disks are different, because they don't use "vbus, target, lun" addressing. If the above sequence doesn't work, then you may have to reboot...

PowerPath doesn't need to be "restarted". It is designed to allow disk changes on the fly. You check to see if it's running by going:

ps -ef | grep -i emc

and you should see about 5 daemons with "emcp" names...

[Rita C Workman · Accepted solution]

Stuart has really given you a great answer, but I'm just going expound on one point.

powermt check

I love this little utility....when you change things run this one. It will ask you about dead connections and cleaning them up and that is exactly what you want..so say Y or say A (yes to all). Then as Stuart says run to hold your configuration:
powermt config
powermt save

And don't be surprised if you have to do it more than once....I like to run this, run ioscan again then recheck my powerpath by running the powermt check again. If it runs clean...I run the config/save and it's done.
But sometimes it takes a little for everything to clean up neat and tidy.

Just a thought,
Rita

[Stuart Abramson]

The full sequence is:

ioscan -Cdisk
insf -eCdisk
powermt display paths
powermt restore
powermt check
powermt config
powermt save
powermt display paths

[Mark Greene_1]

You shouldn't have to reboot, it's not a windows box. Don't do a powermt restore if you have dead paths due to volumes being removed on the SAN side, you'll either hang the system or end-up not being able to see any of the volumes.

What may have to happen is the SAN admin will have to take the host out of the storage group--but not the volumes--then put it back. This will reinitiate the connectivity from the SAN side. Then do ioscan, insf, powermt check, powermt config, powermt display dev=all and everything should be all connected and visible.

The actual powerpath process is /etc/emcpdaemon. The start-up should be in /sbin/init.d/emcp.

Once you have it all back, you'll want to do a powermt save [file] with the file you are saving to somewhere else then /etc so one, it doesn't get overwritten if something like this happens again, and two, so you can explicitly restore from it if something like this happens again.


mark

[Silver_1]

Thanks Guys.