Luns visibility Issue

[Prasanth V Aravind]

Hi all,



We have reinstalled one of old servers (Model : rp7410 OS :- hpux 11.11) in our datacenter, want to use for testing environment.

Our requirement is 5TB san luns in this server. But we are facing luns visibility issue.

Only 8 luns are visible after allocating 170 luns from IBM DS8k storage.

I am new in storage, so not able to identify that where ths issue is ?

Is it fc card issue ?

Prasanth V Aravindakshan

[R.K. #]

Hi Prasanth,

Have a look on this:

http://docs.hp.com/en/6054/LVM_Limits_White_Paper_V4.pdf

and the attachment.

Redgs..

[Michal Kapalka (mikap)]

hi,

ask the SAN administrator, if he presented all LUNs to your server.

mikap

[Viveki]

Please see
http://forums11.itrc.hp.com/service/forums/questionanswer.do?admit=109447626+1255710355844+28353475&threadId=1117692

[Michael Steele_2]

Hi

Your responsibilty stops after running:

inst -f (* once or twice *)
inscan (* verify new device paths *)

If you've recreated new device paths with insf and you still can see them in ioscan then send it back to the SAN administrator.

Questions: Are you running any multipath application? What does this say.

Questions: Can you see the lun ids with a disk array command? Or the disk array id?

[Michael Steele_2]

Sorry, that's

insf -e
ioscan -fnk

[Johnson Punniyalingam]

>>I am new in storage, so not able to identify that where ths issue is ?<<<

In this case , Please take advice from the Storage Team support

>>Is it fc card issue ? <<<

No if its fc card issue , you will not be able to seen nonce of the newly presented LUNS.

Chance could could be :-

once LUNS has been presented have you issued the below commands

# ioscan -fn
# insf -e
# ioscan -fnC disk

rgds,
Johnson

[Prasanth V Aravind]

I feel its a fc card issue because i got a error on eventlog

" Adapter at hardware path 1/0/2/0/0 : Fibre Channel driver has gone into non-participating mode"

...

what exactly mean by non-participating mode ?

[Viveki]

Hi Prasanth,

It is not an FC card issue. The message might occured while you rebooted the server and the nogotiation is going on between the HBA and the other side device. This is related to the driver of the FC card. This event is known to be ignorable. There is patch to avoid this.

The patch "s700_800 11.11 Tachyon TL Fibre Channel Driver Patch (PHSS_31326 )" should stop these events in future.

By the way, did you check the storage settings? Or you can log a support case with HP if you still suspects an issue with the system. I feel the system or the card doesnt have any issues as you are able to see 8 LUNs

[Michael Steele_2]

Hi

Checking the HBA may or may not be the solution. Checking the firmware levels; They must be the same on each, is certainly basic and necessary.

Quick specs for the rp7410 says :

PCI-X Dual Channel 2Gb Fibre Channel HBA A6826A

Lastest firmware: EFI 1.50 / RISC 3.3.171
11 May 2007

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareIndex.jsp?lang=en&cc=us&prodNameId=331475&prodTypeId=329290&prodSeriesId=331473&swLang=13&taskId=135&swEnvOID=54

PCI 2Gb single port fibre channel adapter A6795A

1.37 (17 Jul 2007)

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=US&swItem=ux-51838-2&prodTypeId=329290&prodSeriesId=432322

Check current with

ioscan -funC fc

fcmsutil /dev/fc## vpd

[Viveki]

Michael is correct .....

Checking the FW of the HBAs is required. But a few points

How many FC cards are used to access the storage (paths)? If you have 2 fc cards and both having the same issue, you need to look beyond the HBAs,

Check the HBA card details from fcmsutil output and update the FW accordingly. you can use the product number to search in Hp.com to get the part details and click on download drivers and softwares to down load the drivers. The product number can be obtained from the fcmsutil command output.

[Michael Steele_2]

It is not an FC card issue.

Tachyon TL Fibre Channel

the system or the card doesnt have any issues

[Prasanth V Aravind]

These are the luns we allocated from storage.

The storage documentation says that, Due to limitations in the host operating system, LUNs greater than x'3FFF' are not supported. When you create or assign LUNs and volumes, only LUN and volume IDs less than x'3FFF' are supported. This limits the maximum number of volumes that are allowed for HP host types to 16384.

Luns Allocated:-
108B,118B,10FB,11FB,6834,6934,6835,6935,6836,6936,6837,6937,6838,6938,6839,6939,683A,693A,683B,693B,683C,693C,683D,693D,683E,693E,683F,693F,6840,6940,6841,6941,6842,6942,6843,6943,6844,6944,6845,6945,6846,6946,6847,6947,6848,6948,6849,6949,684A,694A,684B,694B,684C,694C,684D,694D,684E,694E,684F,694F,108A,118A,10FA,11FA,6818,6918,6819,6919,681A,691A,681B,691B,681C,691C,681D,691D,681E,691E,681F,691F,6820,6920,6821,6921,6822,6922,6823,6923,6824,6924,6825,6925,6826,6926,6827,6927,6828,6928,6829,6929,682A,692A,682B,692B,682C,692C,682D,692D,682E,692E,682F,692F,6830,6930,6831,6931,6832,6932,6833,6933,66FF,67FF,6800,6900,6801,6901,6802,6902,6803,6903,6804,6904,6805,6905,6806,6906,6807,6907,6808,6908,6809,6909,680A,690A,680B,690B,680C,690C,680D,690D,680E,690E,680F,690F,6810,6910,6811,6911,6812,6912,6813,6913,6814,6914,6815,6915,6816,6916,6817,6917

Luns Visible:-
108B,118B,10FB,11FB,108A,118A,10FA,11FA,


Here we are able to see the luns with lun_ids < 3FFF.

any fix available for this ?

[Michael Steele_2]

Hi

"...The storage documentation says that, Due to limitations in the host operating system, LUNs greater than x'3FFF' are not supported..."

This is between IBM and HP and IBM states not supported. You'll have to verify with IBM as well as HP, but, start a Case with HP and get it documented. I'm sure they know about.

[Viveki]

Hi Prasanth,

You done a good job digging this out.

I dont think there is a fis avaialbel for this. You need to change the LUN numbers. If it is possible, need to recreated them with supported LUN Ids

[Prasanth V Aravind]

I had started a Case with HP. But i feel, i will get faster response form forum.

let hp to debugg on this further.

Thanks for your support & suggestions ... :)

[Suraj K Sankari]

Hi,

One thing you can try remove all lun's and recreate it and second try to assign into some different server then you came to knwo where will be the proble.

Suraj

[Viveki]

Or just create 2-3 test LUNs with supported IDs and check. If they are able to detect, the issue is confirmed

[Prasanth V Aravind]

I have allocated the same luns to another server & having same issue there also.

If i am allocating new luns with supported LUNID, it getting detect there.

Also i got the confirmation from HP team that, hpux have limitation of 16384 luns &

THERE IS NO FIX AVAILABLE for this..

[Duncan Edmonstone]

Prasanth,

Are you saying this one system has more than 16384 LUNs attached to it?

Or merely that the "LUN number" from the IBM DS8k is higher than 16384?

The reason I ask, was that I'm pretty sure the maximum number of attached LUNs for 11.11 was actually way below 16384, which makes me think what we are actually referring to here is the "LUN number" which is a function of the IBM system - on any other disk array its usually possible to restart LUN numbering on a per host basis - i.e. For each host the disk array restarts the LUN numbering back at zero - that's ceratinly possible with HP and EMC disk arrays - I can't speak for IBM ones though - never used them...

So if its just a "LUN numbering" issue and not a "total number of LUNs" issue, I'd go back to IBM and request an alternative solution (on the basis that their competitors disk arrays don't have this limitation)

HTH

Duncan

[Viveki]

Duncan,

This is a LUN numbering issue. If assigning LUN IDs < 3FFF, system is able detect. And ofcourse 3FFF is a huge value (16383). I don't think anybody requires more than 16383 LUNs in his system ;). So this cannot be called a limitaion to the HP-UX.

[Prasanth V Aravind]

I have allocated luns which having smaller lunids & issue got resolved

THanks for support