Only single path showing from storage

[Syedfurquan]

Hello Expert,

I have rp5470 server with hp-ux 11.11 connected with MSA2000 storage directly having its own Brocade switch. The zoning has been done between the server & Storage i think as it is connected directly..

When I check the ioscan -fnC fc O/P it gives following O/P

# ioscan -fnC fc
Class     I  H/W Path  Driver S/W State   H/W Type     Description
=================================================================
fc        0  0/3/0/0   td   CLAIMED     INTERFACE    HP Tachyon XL2 Fibre Channel Mass Storage Adapter
                      /dev/td0
fc        1  0/9/0/0   td   CLAIMED     INTERFACE    HP Tachyon XL2 Fibre Channel Mass Storage Adapter
                      /dev/td1

But when I check the ioscan -fnC disk it shows disk only from path:

0/9/0/0.1.0.

I would like to know why the disk from other path is not showing. Is it because of zoning not done between other fc or something else.

How to verify this and make the disk availaible from other fc path as well.

Rgards,

[Greg Tinker]

Syedfurguan,

good question - you either have a bad path, or failure in your array controller - I assume something simple like a zone issue.

Here is how to tell.

fcmsutil /dev/tdo get_remote_all

look to see if your tachyon driver can see remote storage array ports - you will see the WWPN of the target - this will prove your SAN is good and that the problem is on the array.

if you can see the WWPN (use td1 to see a good example) then you may just have a lun masking problem on the array.

Regards,

G Tinker

[Syedfurquan]

Hi Greg,

Thanks for the reply.

Sorry it was typo mistake in my last post when td0 shows me fc driver not loaded.

 I run the commands and the O/P are:

# fcmsutil /dev/td1 get_remote_all
*************************************************************************
*****          IOCTL functions available to the TD driver           *****
*************************************************************************

fcmsutil < TD device_file >
         < TD device_file >  echo  <remote N_Port_id> [data_size] [count]
         < TD device_file >  test  <remote N_Port_id> [data_size] [count]
         < TD device_file >  rls   <remote N_Port_id>
         < TD device_file >  [-f]  lb  plm|crpat|cjtpat [count]
         < TD device_file >  read_cr
         < TD device_file >  read  <offset> [pci]
         < TD device_file >  write <offset> <value> [pci]
         < TD device_file >  vpd
         < TD device_file >  get   fabric
         < TD device_file >  get   remote {all | <remote N_Portid>}
         < TD device_file >  [-k] ns_query_ports
         < TD device_file >  reset
         < TD device_file >  [-f] bdr <target device_file>
         < TD device_file >  replace_dsk <remote N_Port_id>
         < TD device_file >  stat [-s]
         < TD device_file >  clear_stat
         < TD device_file >  nsstat
         < TD device_file >  clear_nsstat
         < TD device_file >  devstat {all | <remote N_Port_id>}
         < TD device_file >  clear_devstat {all | <remote N_Port_id>}
         < TD device file >  enable
         < TD device file >  [-f] disable
Note: <remote-N-port-ID> can be substituted with -w <wwn>
      or -l <loop_id>(only in PRIVATE_LOOP) for all options
      that take <remote-N-port-ID> as an argument.

# fcmsutil /dev/td0 get_remote_all
*************************************************************************
*****          IOCTL functions available to the TD driver           *****
*************************************************************************

fcmsutil < TD device_file >
         < TD device_file >  echo  <remote N_Port_id> [data_size] [count]
         < TD device_file >  test  <remote N_Port_id> [data_size] [count]
         < TD device_file >  rls   <remote N_Port_id>
         < TD device_file >  [-f]  lb  plm|crpat|cjtpat [count]
         < TD device_file >  read_cr
         < TD device_file >  read  <offset> [pci]
         < TD device_file >  write <offset> <value> [pci]
         < TD device_file >  vpd
         < TD device_file >  get   fabric
         < TD device_file >  get   remote {all | <remote N_Portid>}
         < TD device_file >  [-k] ns_query_ports
         < TD device_file >  reset
         < TD device_file >  [-f] bdr <target device_file>
         < TD device_file >  replace_dsk <remote N_Port_id>
         < TD device_file >  stat [-s]
         < TD device_file >  clear_stat
         < TD device_file >  nsstat
         < TD device_file >  clear_nsstat
         < TD device_file >  devstat {all | <remote N_Port_id>}
         < TD device_file >  clear_devstat {all | <remote N_Port_id>}
         < TD device file >  enable
         < TD device file >  [-f] disable
Note: <remote-N-port-ID> can be substituted with -w <wwn>
      or -l <loop_id>(only in PRIVATE_LOOP) for all options
      that take <remote-N-port-ID> as an argument.

It looks both of the fc are fine.I checked from the fcmsutil as well and it shows me driver state as online.

Regards,

[Greg Tinker]

Sorry - I could not remember if it was a "_" or just a space " " :)

remember, its a zero "0" not the letter "o"

Example:

fcmsutil /dev/td0 get remote all

                    Target N_Port_id is = 0x651213

                           Target state = DVS_UNOPENED <-- this means you have no bindings in use... 

                          Symbolic Name =

                              Port Type = N_PORT

   SLER Capable (supports Retry & TRID) = YES

            Target Port World Wide Name = 0x100000e002220789 <-- This is what you are looking for, your WWPN will be different

            Target Node World Wide Name = 0x100000e002020789

[Syedfurquan]

Hi,

Requested O/P

fcmsutil /dev/td0 get remote all

                    Target N_Port_id is = 0x0000e4
                       Target Loop_id is = 2
                           Target state = DVS_UNOPENED
                          Symbolic Name =
                              Port Type = NL_PORT
   SLER Capable (supports Retry & TRID) = YES
            Target Port World Wide Name = 0x100000e002027a59
            Target Node World Wide Name = 0x100000e002027a47

Common Service parameters (all values shown in hex):
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
               Version_Hi: 20                     Version_Low: 20
                BB_Credit: 0                             Ciro: 1
               Alt_BBcred: 1                             Rxsz: 400
                   Conseq: ff                       RO_Bitmap: fe
               Resolution: 0                        EDTOV(ms): 7d0

Vendor version (in hex) 0       0       0       0
-------------------------------------------------------------

# fcmsutil /dev/td1 get remote all

                    Target N_Port_id is = 0x010000
                           Target state = DVS_READY
                          Symbolic Name =
                              Port Type = N_PORT
   SLER Capable (supports Retry & TRID) = NO
            Target Port World Wide Name = 0x500805f300130ce1
            Target Node World Wide Name = 0x500805f300130ce0

Common Service parameters (all values shown in hex):
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
               Version_Hi: 20                     Version_Low: 9
                BB_Credit: 0                             Ciro: 1
               Alt_BBcred: 1                             Rxsz: 800
                   Conseq: ff                       RO_Bitmap: fe
               Resolution: 0                        EDTOV(ms): 0

Vendor version (in hex) 51435143        31303020        400     0
-------------------------------------------------------------

Now how to resolve this issue?

Regards,

[Greg Tinker]

You have a zoning problem.

HBA Driver instance td1 is zoned to an HP P2000 Storage Array controller (just guessing by the WWPN 500805) but td0 does not see the same storage array - it appears to see a Tape Device - but I do not have all vendor/product WWPN's memorized  :)

In short, you have a zoning problem - Speak to your SAN Administrator to correct zoning so that the two HBA's can see the same storage array (both controllers).

The SAN admin will need the two WWPN from your HBA's

command is

fcmsutil /dev/td0

fcmsutil /dev/td1

look for the WWPN of the card, and provide that to the SAN admin he/she must then add the HBA to the Zone containing the WWPN of the P2000 Host Ports... 

Then ioscan -fn, followd by insf -e, then you either have to have a multipath driver (autopath, hdlm, securepath, etc...) or just use native LVM multipathing.  -- HPUX 11.31 (11iv3) automates all of this.

[Syedfurquan]

Hi Greig,

Thanks for your analysis and support.

I am new to this site recently taken handover and have to look into all the issue related to hp-ux.There is no separate SAN zoning between the server and MSA as there is no separate physical SAN.

I can see one switch connected on the back of MSA and both the fc are connected there.

Now how can I login onto that switch and do the required configuration.

 I don't have much expertise on switch end.

Appreciate your support if you can provide me the steps to resolve this issue.

Is there chance that the fc0 is faulty and because of that only it's not detecting?

Regards,

[ChrisTinker · Accepted solution]

Note hat TD0 is able to see devices, not the device you want.  We can reasonably deduce that the HBA is functioning.  More to the point, when you issue the “get remote all”, the HBA driver sends a NS_QUERY_PORTS() function call to the switch where the switch’s Name Server (Not DNS) responds with the Target FCID’s for which the Iniator is allowed to talk to.  After this, a report_lun() is issued and thus we see the devices from whatever targets are found.  To this end, all support that the HBA is operating correctly.  This is why the zoning is suspect – it could be that the port on the array that we are supposed to see is bad/failed.

TD0 -------    FABRIC A   ----- TARGET NOT FOUND

               FABRIC A   ----- TARGET 0x100000e002027a59

TD1 -------    FABRIC B   ----- TARGET 0x500805f300130ce1 

               FABRIC B   ----- TARGET ???  Why do we not see 0x100000e??

How to login to the switch:  

Depends on switch type.  You would need the user/pass and I would recommend using the web interface or the GUI API.  Given your statement regarding “not much experience with switches”, I would like to say careful, as the switch controls all access to your storage, you risk not only breaking this host, but ripping all storage away from all hosts in the fabric.

I advise placing a call to HP support  800-633-3600 (contract support)  in order to have someone with more experience guide you.  You are also welcome to download the user guide for the matching switch that you have which would be able to take you through the steps.

The storage design guide is @ http://h20000.www2.hp.com/bc/docs/support/SupportManual/c00403562/c00403562.pdf

[Syedfurquan]

Hi Chris,

Thanks alot for explaining do deeply. As mentioned in my previous post as well that the switch is on the back of this storage only and this storage is low end and connected only with the present server and all disks are populated to this server.

Now since this switch is not connected on the network then how can I login onto this switch.

Regards

[karthikbalu]

May be this can looked in other way. On dmesg you might be seeing some Lunpath errors....if it populates then issue the below command to replace the path with actual one.

scsimgr replace_wwid -C lunpath -I 5   ###You have to change the value '5' according to the lunpath error you received.