HP-UX

From the development system, please post the output from:

Code:

strings /etc/lvmtab

It is almost as if the development server is seeing the Live servers discs c*t0d* and not seeing its own discs c*t1d*. This is pure guesswork because we don't know the correct LVM configuration or hardware configuration for the development server.


Ps. Just seen post #9. /etc/lvmconf is a directory not a file.

Code:

LUNHINGA:[/]#strings /etc/lvmtab
/dev/vg00
/dev/dsk/c2t1d0s2
/dev/dsk/c2t0d0s2
/dev/vgdvqatr
/dev/dsk/c8t0d4
/dev/dsk/c9t0d4
/dev/dsk/c10t0d4
/dev/dsk/c11t0d4
/dev/vglunsap
/dev/dsk/c8t0d5
/dev/dsk/c9t0d5
/dev/dsk/c10t0d5
/dev/dsk/c11t0d5
/dev/vgr3dev0
/dev/dsk/c8t0d1
/dev/dsk/c9t0d1
/dev/dsk/c10t0d1
/dev/dsk/c11t0d1
/dev/vgr3qas0
/dev/dsk/c8t0d2
/dev/dsk/c9t0d2
/dev/dsk/c10t0d2
/dev/dsk/c11t0d2
/dev/vgr3trn0
/dev/dsk/c8t0d3
/dev/dsk/c9t0d3
/dev/dsk/c10t0d3
/dev/dsk/c11t0d3
/dev/vgr3trn1
/dev/dsk/c8t1d0
/dev/dsk/c9t1d0
/dev/dsk/c10t1d0
/dev/dsk/c11t1d0
/dev/vgr3dev1
/dev/dsk/c8t0d6
/dev/dsk/c9t0d6
/dev/dsk/c10t0d6
/dev/dsk/c11t0d6
/dev/vgr3qas1
/dev/dsk/c8t1d1
/dev/dsk/c9t1d1
/dev/dsk/c10t1d1
/dev/dsk/c11t1d1

I dont know what you think methyl, but seems we have duals paths...
If you happen to not declare properly the alternates you do get this sort of issues ...

The contents of /etc/lvmtab is an OMG moment.

My impression (without anywhere near all the facts) is that this is a physical (or logical?) cabling error. The development server is seeing the live discs not the development discs. It now has a corrupt /etc/lvmtab following the ill-advised vgchange -a y in post #1 which is a bit like running vgscan out of context.
Hopefully the O/P should be able to recover /etc/lvmtab and the entire contents of /etc/lvmconf from backup if requested by someone who asks.

I don't know about "vbe" but I am long past the point where I would need physical access to ALL the servers involved (not just the two mentioned).

Time to log a formal call with HP against your HP Maintenance Contract. There is enough evidence in this thread to give them a head start.

Based on what we have seen the development server thankfully cannot mount any partition on the live discs. Personally I would shut down the development server and physically isolate it from the live server's discs pending repair. IMHO, typing any more LVM commands on the development server is foolish.

I agree methyl, since we still dont know how disks are connected but there is 2 FC, this looks very much like what you describe, and it can happen with wrong zoning update...

Check both fc's with

Code:

fcmsutil /dev/td0 
fcmsutil /dev/td1

Check for Driver status (should be online) and get both cards WWN.
With WWN information, connect to SAN switch and check zoning.

Does syslog indicate a path loss or any disk subsystem errors ?

Dear Mazen,

I think that you have EVA4000 storage? So probably half of SAN paths are lost, and volume groups are not able to get activated without quorum,
can you post the output of #strings /etc/lvmtab to see the structure of volume groups and advise if it is a MC/SG environment?

Thanks
Francis

---------- Post updated at 01:39 PM ---------- Previous update was at 01:38 PM ----------

it just noticed the remaining part of posts