topic Re: LVM command to recognize failed emc pvlink in Operating System - HP-UX

LVM command to recognize failed emc pvlink

Unix Team_9 — Mon, 02 Mar 2009 04:13:35 GMT

Hi,

Can any one specify the command for recognizing the faild emc pvlink in hp-ux

Re: LVM command to recognize failed emc pvlink

Jozef_Novak — Mon, 02 Mar 2009 12:02:49 GMT

Hello,

if a particular PV link has failed, it should be displayed as NO_HW in ioscan output.

# ioscan -fnC disk

search for NO_HW state disks, then locate respective raw device file for each failed disk. Should look like this:

/dev/rdsk/cXtYdZ

If you have EMC command line interface installed, you can run the following command to get the device id to which the failed PVlink belongs:

# symdev -sid list | grep

e.g.

# symdev -sid 800 list | grep c1t2d3

If you are using LVM, you can check output of vgdisplay to find out which volume group that disk belongs to:

# vgdisplay -v

Hope this helps,
J.

Re: LVM command to recognize failed emc pvlink

Shahul — Mon, 02 Mar 2009 12:10:00 GMT

Hi,

#ioscan -fnC disk |grep NO should tell you which hardware path is failed. If you want to know the EMC LUN id, then.

powermt display dev=all |grep cXtXdX (provided you have EMC powerpath license).

symdev show will give you that particular device details.

Good luck
Shahul

Re: LVM command to recognize failed emc pvlink

Bill Costigan — Tue, 03 Mar 2009 00:50:56 GMT

If powerpath is being used, you will not see the NO_HW when you do the ioscan.

powerpath will make the downed path look good to ioscan by redirecting the I/O to a good path.

If powerpath is being used, you'll need to use the 'powermt display dev=..' command.

Re: LVM command to recognize failed emc pvlink

Jozef_Novak — Tue, 03 Mar 2009 09:48:20 GMT

Bill,

I guess the question was about PVlinks, which clearly tells which multipathing method is in use.

Using PVlinks does not make much sense when PowerPath is available.

J.

Re: LVM command to recognize failed emc pvlink

Steve Post — Tue, 03 Mar 2009 13:25:19 GMT

"emc pvlinks" could be a few different things. There are different EMC systems.

I'll assume for a second you are using powerpath, with disks hooked up over multiple paths using two fibre switches.

Run man page on powermt.
If you run "powermt display every=5" you will see if any disks (correction: disk PATHS) are dead. Perhaps they are NOT dead. Perhaps it is a fibre cable, or something wrong on EMC? Go into Navisphere gui interface over the web. If you log in there and there is a problem it will tell you. You'll have an icon on upper right side of the web page with a the word "WARNING" on it.

Meanwhile......
Let's say you did in fact lose a disk and got it fixed. The system might not see it. When the system was first set up here, we ran a test failure. One thing we noticed was a 5 minute delay on the system recognizing the disk paths.

I had to run this....(LOOK AT THE MAN PAGE)
powermt check
powermt config
powermt save
powermt set config=co

These commands might not apply to you. You might not even have the powermt command. The command setting config=co might be the LAST thing you want to do. I was told to do this with my Clariion CX300.

.....I have a much better idea. Get with EMC.

Re: LVM command to recognize failed emc pvlink

Unix Team_9 — Thu, 09 Apr 2009 18:52:14 GMT

Thanks