SAN Disks not appearing

[Adel B]

We have an itanium rx4640 HP-UX server. The server has 2 internal disks and several SAN disks using an EVA san storage system. the server is connected to the SAN using 2 HBA fiber channel cards connected to a SAN switch. After a system crash we can no more see the SAN disks. I've run an IO scan which doesn't show the disks but it clearly shows both HBA fc cards and they look normal and claimed. So I used the fcmsutils to troubleshoot the cards. Every thing looks fine except that both of the cards show "not initialized". I don't know if that is related to the problem or not. So I reset both cards using fcmsutils. this didn't solve the problem neither. I looked on the back of the server and both HBA cards show the PWR LED as steady green. However the other LED on the cards is off, which should show also steady grean if activity exists. Any ideas to resolve this issue guys?

[SoorajCleris]

Hi ,

Which is your OS version ?
Are you sure your EVA is up and running and your Fabrc is fine?

is there any other Storage LUNs mapped to this server?

Is there any other server using this EVA which is working fine ?

Could you please paste the output of the ioscan -fun and ioscan -fn command output???

Regards,
Sooraj

[Adel B]

OS : HP-UX 11.23
The EVA is up and running because it's also connected to other servers which don't have any problem.
Also the EVA is operating normally whithout warnings nor errors.
All Luns mapped to this server are not showing.
It's a bit hard for me now to paste the ioscan output. but I can answer if you have a specific question.

[Michael Steele_2]

a) insf -e (* to update the new device paths in the kernel's ioconfig table *)
b) insf -e (* to be sure *)
c) ioscan -funC (* to read the new updates and current device paths *)

[Adel B]

insf -e Already done twice...No luck

[SoorajCleris]

Probably your card must be the culprit .......

Somecards running with old firmware/drivers may show this kind of behaviour ( there are some known issues for some cards.)

Is there any chance to remove the FC cable and reinsert it ????

Regards,
Sooraj

[chris huys_4 · Accepted solution]

Hi,

Sanswitchports of the sanswitches, the server fc hba's are connected to, are probably in disabled state.

Enable the sanswitchports of the sanswitches, i.e if its a brocade sanswitch, telnet ; portenable , and the server fc hba's should be able to do a fabric port login and consequently a ioscan -fn should be able to see back the eva luns.

Also cut&paste the output of fcmsutil /dev/# next time..

Greetz,
Chris

[Adel B]

I did remove the cable and re-inserted it. but nothing improved.

Chris, the SAN switch was working normally and no-one changed any configurations to it. So why could the ports be disabled?

Anyway I'll do what you suggested tomorrow to see if that would help the situation.

[Leonard C.]

1) ioscan
2) ioscan -fun (send output)
3) Do you see fcd0 or fcd1 in /dev ?
4) What's the EVA model?

[Sagar Sirdesai]

Before you run the ioscan of the disk .
You run the ioscan for the controllers.
ioscan -fnCctl.

If this shows any output then the server can see that EVA.
Then run the ioscan for the disks

If no controllers are seen from ioscan output then there is some issues with the zoning at the SAN Swicth Side.


Also let us know what Multipath is been used

Sagar

[Adel B]

The San ports were disabled as you said Chris. But now we have another problem the device files names changed which prevents the volume group from activating.
they used to be c12t0d2 and c12t0d3 but now they are c12t0do and c12t0d1. How can I fix this?

[Sagar Sirdesai]

Hi
vgexport the volume groups with the map files.
Then vgimport it with the correct device files.

You do not want to specify the device files then while you vgexport the volume groups use the -s option.

Then again while vgimporting again use the -s option

Sagar

[Adel B]

Can you please give me more detail on how to vgexport and import please

[Michael Steele_2]

Hi

What version of hp-ux are you using? The procedure below if for 11.23 and earlier. Check to see if you are using 11.31.

This is a complicated task but there are several ways to verify your work.

a) vgcfgrestore -f /etc/lvmconf -l

What you should see are the old devices

b) vgscan -s -a -v -p

-p for preview
-a for search every controller
-v for verbose
-s for sharable pvs - causes vgscan to read the lvm header of every pv, note the vg written in the lvm header, populate /etc/lvmtab by this info.

c) vgexport -p -v -s -m mapfile /dev/vg
-p preview
-v verbose
-s sharable - update the pv lvm header by vg belonging to the pv
-m hp-ux ascii file, used to rebuilt the vg.

d) vgimport -s -p -m mapfile /dev/vg
-p preview
-m mapfile
-s sharable - add pv's based upon lvm header

here is a good thread.

http://forums13.itrc.hp.com/service/forums/questionanswer.do?admit=109447627+1270558959757+28353475&threadId=1287624

[Adel B]

Thanks a lot guys the problem was solved by enabling the San switch ports which were disabled.

Then exporting and importing the VGs.

[Adel B]

San Switch Enable

Export/import VGs