Boot disk fails

[Vishu]

Hi Masters,

I have two query. i have HP-UX 11.00 with two disks in Vg00 mirrored with each other. Just want to know if the primary boot disk, from which my server booted, fails then will my server be up on the alternate one or it will panic?

and, if it remains up, then can i replace the disk with the new one, if it is hot swap? and after that server will come back to my primary disk?

please input on above.

Thanks

[James R. Ferguson · Accepted solution]

Hi:

> i have HP-UX 11.00 with two disks in Vg00 mirrored with each other. Just want to know if the primary boot disk, from which my server booted, fails then will my server be up on the alternate one or it will panic?

Toss a coin. Depending on the nature of the failure, you may either remain running or you may have to reboot. LVM MirrorDisk/UX is a high-availability feature not a fail-safe one for the boot disk.

> if it remains up, then can i replace the disk with the new one, if it is hot swap? and after that server will come back to my primary disk?

Without knowing the hardware, it isn't possible to say if the disk is a hot-swappable one. Given that you replace the disk and sucessfully configure it again as a vg00 disk, and given that your primary boot path remains unchanged, then "yes" to your question.

The 11.0 release is out-of-support, but disk replacement procedures can be found here:

http://docs.hp.com/en/5991-1236/When_Good_Disks_Go_Bad_WP.pdf

Of course, some of the newer features won't be of any help to you.

Regards!

...JRF...

[Vishu]

Thanks JRF for promptly replying.

But,

***Toss a coin. Depending on the nature of the failure, you may either remain running or you may have to reboot. LVM MirrorDisk/UX is a high-availability feature not a fail-safe one for the boot disk.****

I am not clear in this as it again put me in same state, where i was...whether it will panic or running. okay i give you that my disk is totally failed and is unavailable...or you can say someone pulled out my primary hot swap disk, then what will happen..

[Bill Hassell]

> Just want to know if the primary boot disk, from which my server booted, fails then will my server be up on the alternate one or it will panic?

LVM mirroring is realtime and the primary disk will be used until it fails. The driver will try the mirror and if it works, further activities will continue on the mirror.

Now whether the system will boot up on the mirror in case the primary is gone or completely failed is critically dependent on whether mkboot was correctly run on both disks when mirroring was first setup.

> can i replace the disk with the new one, if it is hot swap? and after that server will come back to my primary disk?

No, and that would be a bad thing if it did. The replacement disk contains nothing but trash so all the extents on the new primary will be marked stale. Once you replace the disk, you'll need to remirror back to the primary. After that, a reboot will use the primary disk. If you don't know if the server was setup correctly, I would definitely test this capability during a maintenance window. If you don't, the next disk failure will confirm configuration -- with possibly dire consequences.

[Raj D.]

Ranju,

>Just want to know if the primary boot disk, from which my server booted,

-This will give the hardware path of the primary boot disk (The disk system is booted from) :
# echo boot_string/s | adb /stand/vmunix /dev/kmem



- To check both the boot disk :
# lvlnboot -v vg00


- In case of server crash , whether your server will comeup with Alternate disk: ?
As mentioned above, it depends how your server mirroring is done, And after mirroring it is a good idea to check the system boot from the Alt disk.
- Sometime you have to choose manually the alternate disk , at PDC to boot from the alternate disk , if setboot is not defined properly to ALT path, and also sometime it will complain about quorum (As first disk is failed) , so you have to use at PDC : hpux -lq ( less quorum). And the system will boot up.

- Once the system is up , Re-mirror the failed disk . (Mirroing steps :on Page#118 http://docs.hp.com/en/5992-6576/5992-6576.pdf )
And thats it.



- Next step boot it back from the primary disk (recently replaced one).


Hth,
Raj.




* Dont forget to assign points, before closing the thread. Cheers.

[Raj D.]

Ranju,
Root disk mirror steps for PA_RISC server:
Earlier page number mentioned is not correct for PA_RISC 11.00.

Correct document attached:
cheers,
Raj.

[Raj D.]

Ranju,

>and, if it remains up, then can i replace the disk with the new one, if it is hot swap?

-Hotswap disks are made for replacing online.



Once you pulled the primary disk from the hotswap bay, the mirrored alternate disk will take care of the system. System will assume the primary disk has failed.

Then follow the below steps to remirror it.
- I have seen few cases , pulling the disk in a highly loaded server, corrupts some files. Once everything was done at midnight after disk replacement and mirroring , got a call at 2am that telnet not working. That was very strange , and found the telnet and ftp program binary was corupted. So just a precaution.





Steps to mirror the replaced disk:
------------------------------------------
Once the failed disk has been replaced : (assuming the disk is c1t1d0):

- vgcfgrestore -n /dev/vg00 /dev/rdsk/c1t1d0

- mkboot /dev/rdsk/c1t1d0
- mkboot -a "hpux -lq (;0) /stand/vmunix" /dev/rdsk/c1t1d0

- vgchange -a y /dev/vg00
- lvlnboot -Rv /dev/vg00

- vgsync /dev/vg00 # it will take around 1 to 2hr to complete the sync depending on disk size.
- Done.
---------------------------


hth,
Raj.

[subodhbagade]

Hi,

(1) If root is mirror then try to boot the server from mirror one.

(2) If the disk is hot swappable then you can replace it online.

Please refer the attachment for your reference.

(3) Once primary bootable disk has been replaced theen try to boot from primary one.


Regards,
Subodh.