Operating System - HP-UX
1753365 Members
6319 Online
108792 Solutions
New Discussion

Disk MultiPath and VG Failure

 
SOLVED
Go to solution
Binubaby
Frequent Advisor

Disk MultiPath and VG Failure

Hello All

 

Please see below issues and advise (HP-UX 11.31 2 Node cluster)

Noticed that half of the path to disk dev/rdsk/c9t0d1  are claimed and active, But still the path to the disk has FAILED

 

 Cluster_Lock_LVM:
    VOLUME_GROUP          PHYSICAL_VOLUME       STATUS             
    /dev/vg_lock          /dev/dsk/c9t0d1       down

 

ioscan shows (also other paths) Please find attached

    disk      0  0/0/8/1/0.1.8.0.0.0.1    sdisk   NO_HW       DEVICE       HP      HSV210
    /dev/dsk/c9t0d1   /dev/rdsk/c9t0d1  
   

vgdisplay output:

vgdisplay -v /dev/vg_lock

vgdisplay: Volume group not activated.

vgdisplay: Cannot display volume group "/dev/vg_lock".

 

I have attached ioscan, lvmtab outputs

 

Please advice how to recover the situation.

 

Thanks

BB

 

 

 

 

 

BB
6 REPLIES 6
Patrick Wallek
Honored Contributor

Re: Disk MultiPath and VG Failure

It looks like everything on fibre path 0/0/8/1/ is down.  Check your fibre card, fibre cable, switch, switch port, SAN port, etc.  Anything between that fibre card and the SAN could cause problems.

 

If you just have 2 paths to that device, try activating the VG in low-quorum mode.

 

# vgchange -a y -q n /dev/vg_lock

 

Then try your 'vgdisplay' again.

Binubaby
Frequent Advisor

Re: Disk MultiPath and VG Failure

Thank You Patrik for your time and inputs

 

I will check as advised.

 

Could you please check below queries

Since there are claimed path available to this VG  why not multipath failover not happend to the alternate claimed path??

 

/dev/vg_lock

/dev/dsk/c9t0d1-- NO_HW

/dev/dsk/c11t0d1-- NO_HW

/dev/dsk/c5t0d1 -- NO_HW

/dev/dsk/c7t0d1 -- NO _HW

/dev/dsk/c17t0d1 -- Claimed

/dev/dsk/c19t0d1 -- Claimed

/dev/dsk/c13t0d1 -- Claimed

/dev/dsk/c15t0d1 -- Claimed

 

This could be a reason of configuration issue? If so how to fix this autofailover issue?

 

# vgchange -a y -q n /dev/vg_lock

How to revert back from low quorum mode, to the original when the original path issue will get resolve.

 

Please respond

BB

 

  


 

 

 

 

BB
Torsten.
Acclaimed Contributor

Re: Disk MultiPath and VG Failure

One problem is the use of the legacy device file (/dev/dsk/c9t0d1) instead of the new agile files (e.g. /dev/disk(disk7).No need to "get

out" of quorum mode, it just disables the check for quorum.


Hope this helps!
Regards
Torsten.

__________________________________________________
There are only 10 types of people in the world -
those who understand binary, and those who don't.

__________________________________________________
No support by private messages. Please ask the forum!

If you feel this was helpful please click the KUDOS! thumb below!   
Binubaby
Frequent Advisor

Re: Disk MultiPath and VG Failure

Hello

If I am using legacy paths will the multipath will work incase any of the path fail? or only for agile paths?

Do I need to do a vgextend to all alternate paths to make it work automatically? in this senario

I have checked all posiibilites of a faulty cable/Switch/Zoning/SFP etc. but all looks fine... (Also other paths to the disks are claimed)

So what could be the reason of these path failure??  still I cannot access the disk.

Any Suggestions

 

Thanks

BB

 

BB
Robert_Jewell
Honored Contributor
Solution

Re: Disk MultiPath and VG Failure

> If I am using legacy paths will the multipath will work incase any of the path fail? or only for agile paths?

 

Legacy multipathing is supported through the scsimgr attribute leg_mpath_enable   For example, display current value of leg_mpath_enable for disk17:
    # scsimgr get_attr –D /dev/rdisk/disk17 –a leg_mpath_enable
    SCSI ATTRIBUTES FOR LUN : /dev/rdisk/disk17
    name = leg_mpath_enable
    current = true
    default = true

 

If you find this disabled for your LUN's and confirm that you need to enable this setting then use the scsimgr set_attr command (note that it is enabled by default, so if you found it disabled, someone had to make the decision to change it).

 

You can also use scsimgr to view the status of the LUN paths:

 

    # scsimgr lun_map

 

View LUN health as follows:

 

    # ioscan -P health

 

> Do I need to do a vgextend to all alternate paths to make it work automatically?

If you use the new persisent device file format there is no need to use vgextend as the single device path provides access to each physical path to the LUN.  In reality the old mult-pathing called PV Links or Alternate links has been deprecated by this.   The following links are to good documents explaining the Multi-Pathing found in 11.31

 

http://h20000.www2.hp.com/bc/docs/support/SupportManual/c01906512/c01906512.pdf
http://h20000.www2.hp.com/bc/docs/support/SupportManual/c01915618/c01915618.pdf

 

-Bob

 

NOTE:

Edited 10/19/2011 by Warren_Admin to fix broken document links.

Warren_Admin, HP Enterprise Community Managing Administrator
Any opinions expressed are mine alone and do not represent HP

 

 

----------------
Was this helpful? Like this post by giving me a thumbs up below!
Binubaby
Frequent Advisor

Re: Disk MultiPath and VG Failure

Hi Robert

Thank You so much for the detailed information.

 

BB

BB