Unsolved

This post is more than 5 years old

3 Posts

660

June 19th, 2008 06:00

(Powerpath) Booting into Multipath

All,

We are running an IBM Bladecenter HS21, booting from SAN, using RHEL5.

I was able to install RHEL5 and run the OS with no problems using Single Path mode, however after having the SAN people turn on Multipathing, I was unable to boot the machine.

When booting the machine I now receive an error stating "fsck.ext3: Unable to resolve 'LABEL=/boot'

This is only happening when Multipathing is turned on. Single Path is/was no problem.

341 Posts

June 19th, 2008 07:00

George

Can you give us a little more detail?
- HBA make/model/driver/fcode
- Kernel version
- Array type
- PowerPath version.

Looking at a Emulex manual here for the 8-x driver it states that PowerPath is not supported on Linux hosts booting from EMC storage arrays... could this be the issue?

null

3 Posts

June 19th, 2008 08:00

Conor,
Thanks for the reply.

I have the following information for you:


2 QLogic HBA's attached.
HBA1 = QMI2472, Version 1.24
HBA2 = QMI3472, Version 2.01

When booting through Single Path mode, the LUN is seen and mounted by the HBA2.

RHEL5 Kernel version 5.1.19.6 --- We have gone through all the organizations and are assured that the RHEL5.1 is capable of Multipath Boot From SAN.

Using PowerPath version 5.1.

Hope this helps.

2 Intern

 • 

1.3K Posts

June 19th, 2008 10:00

what you have in fstab for /boot?

if the partition is not labeled then try labeling . (e2label work for me.)

here is how the label configured on my systems's fstab

LABEL=/ / ext3 defaults 1 1
LABEL=/boot /boot ext3 defaults 1 2

3 Posts

June 20th, 2008 11:00

This is my fstab file:


# cat fstab
/dev/VolGroup00/LogVol00 / ext3 defaults 1 1
LABEL=/boot /boot ext3 defaults 1 2
devpts /dev/pts devpts gid=5,mode=620 0 0
tmpfs /dev/shm tmpfs defaults 0 0
proc /proc proc defaults 0 0
sysfs /sys sysfs defaults 0 0
/dev/VolGroup00/LogVolSwap swap swap defaults 0 0


Does anything look out of place? This Blade is booting from SAN.

6 Operator

 • 

5.7K Posts

June 22nd, 2008 00:00

Did you enter the boot LUN in both HBAs ? Both HBAs need to "know" the LUN they're expected to boot from. It's also good for redundancy, so when 1 HBA fails, the host can still boot from the other HBA. And of course you need to make sure both HBAs are zoned properly and have access to the boot LUN (masking on Symmetrix / storage groups in Clariion).

2 Intern

 • 

1.3K Posts

June 22nd, 2008 14:00

what does e2label /dev/sdax returns ?
replace the /dev/sdax with the equivalent one in your case.
No Events found!

Top