Operating System - HP-UX
1745795 Members
3694 Online
108722 Solutions
New Discussion юеВ

Problem with alternate links

 
SOLVED
Go to solution
Beno├оt
Regular Advisor

Problem with alternate links

Hi,

My config is HP9000 rp8400 2xFC Cards / HPUX11 and VA7400.
I've created a new LUN, new VG and new LV.

Problem is that I can't configure (well I don't see) all alternate links.
I should have 4 path to my LUN and I get only two.

--- Physical volumes ---
PV Name /dev/dsk/c7t5d2
PV Name /dev/dsk/c21t5d2 Alternate Link
PV Status available
Total PE 19964
Free PE 0
Autoswitch On

Other path should be on c13 and c24 like it is on other existing VG.
Ioscan doesn't show those path.

How to configure that?

Thanks for help.

Ben
20 REPLIES 20
Jeff Schussele
Honored Contributor

Re: Problem with alternate links

Hi Ben,

As this a switched fabric topology?
If so I'd suspect the zoning is the culprit here.
When zoned properly the ioscan should see the new devices & an insf comannd will add them.

HTH,
Jeff
PERSEVERANCE -- Remember, whatever does not kill you only makes you stronger!
Patrick Wallek
Honored Contributor

Re: Problem with alternate links

Did you vgextend the other paths into the VG?

If not, you need to.
Beno├оt
Regular Advisor

Re: Problem with alternate links

Zoning is fine.
All LUN have 4 PVs (4 path).
For this new LUN (same array) I have only 2 PVs.

I did vgcreate vgTEST /dev/dsk/c7t5d2 then vgextend vgTEST /dev/dsk/c21t5d2 for the alternate link.

Looking on already existing VGs I see that the 2 other path should be c13 and c24.
But a vgextend vgTEST /dev/dsk/c13t5d2 fails.
Patrick Wallek
Honored Contributor

Re: Problem with alternate links

How does it fail?

What happens if you do a 'diskinfo /dev/rdsk/c13t5d2'? Do you see the c13t5d2 and c24t5d2 paths with ioscan?
Beno├оt
Regular Advisor

Re: Problem with alternate links

diskinfo /dev/rdsk/c13t5d2
diskinfo: can't open /dev/rdsk/c13t5d2: Invalid argument

ioscan doesn't show c13t5d2 or c24t5d2.
Patrick Wallek
Honored Contributor

Re: Problem with alternate links

OK then. There's your problem. You need to figure out WHY those paths aren't there.

Do you have Secure Manager installed on the VA7400? If so, did you allow access to the new LUN via all 4 paths via 4 WWN's?

There is something still amiss somewhere.
Beno├оt
Regular Advisor

Re: Problem with alternate links

I didn't configure anything special for my LUN.

I did: armcfg -L 170 -a 78G -g 1 smcva11

That's all.
Florian Heigl (new acc)
Honored Contributor

Re: Problem with alternate links

Benoit,

if there's two less PVs for this LUN then for all others, then there's something not okay, not on LVM, but on either LUN presentation or hardware access.

A few guesses:
- more than 127 LUNs? enable FC_LARGE_CONFIG
- don't show up for the controller (as far as I understand, that's the case)? the storage system hasn't mapped them available to Your host?
- zoning ought to be ok if You see other LUNs over these paths, BUT there's a difference to look at: are Your doing port or wwn zoning? for wwn zoning, You probably need a zone for every LUN.
- ioscan / insf were run incorrectly
- look if /opt/fcms/bin/tdutil [adapter] get remote all is able to see the missing PVs
- look if ioscan -fnd shows them with a wrong driver (i.e. ctl - I once had that error)
- for symmetrix and maybe eva/xp You might need an extra command to rediscover. (syminq, I believe)

Good luck,
Florian
yesterday I stood at the edge. Today I'm one step ahead.
Beno├оt
Regular Advisor

Re: Problem with alternate links

- more than 127 LUNs? enable FC_LARGE_CONFIG

Where do I see that?
Should be ok, all previous config has been done by HP.

- don't show up for the controller (as far as I understand, that's the case)? the storage system hasn't mapped them available to Your host?

There is no mapping/LUN masking.
I know it's bad but I not responsable of that.
So every server on the SAN can see those LUNs

- zoning ought to be ok if You see other LUNs over these paths, BUT there's a difference to look at: are Your doing port or wwn zoning? for wwn zoning, You probably need a zone for every LUN.

This is port zoning, and I'm sure it's ok coz this server can see around 100 LUNs correctly from this array.

- ioscan / insf were run incorrectly

I did: ioscan -fnC disk
I didn't think an insf was necessary (didn't do it)
Is it necessary?

- look if /opt/fcms/bin/tdutil [adapter] get remote all is able to see the missing PVs

I don't know how that works. I'll see...

- look if ioscan -fnd shows them with a wrong driver (i.e. ctl - I once had that error)

No. Should I post result of my ioscan -fn ?
You're going to cry :-)

- for symmetrix and maybe eva/xp You might need an extra command to rediscover. (syminq, I believe)

Here is a VA7400.
Just ioscan is supposed to be necessary