make_recovery

[Nufimi Zainal]

Hi,

I am a bit new at this. Can someone tell me how do I boot up from a make_recovery tape? I need to recover the whole filesystem. Thank you.

Fim

[Rajeev Shukla]

Hi,
Insert the tape in the drive and boot the system.
Interupt the boot sequence by pressing spacebar and then at PDC search for the tape drive say its p0
then boot through po by typing
bo p0
Then u'll get the recovery option and select the desired, it menu based after that.

[Nufimi Zainal]

Hi,

Thanks for the help. However, another question pop up to my mind. If I just want to recovery the whole OS that is the vg00 without any interaction, will my data on another vg be affected? Thanks

Fim

[Michael Steele_2]

In this case, "...I need to recover the whole filesystem...", you don't boot off of the make_recovery tape. Instead you use 'mt' to position the tape and 'tar' to copy the file system off of the tape.

Insert tape.

# mt -t /dev/rmt/0m rew (* Note, 0m *)
# mt -t /dev/rmt/0m rew
# mt -t /dev/rmt/0mn fsf 1

Note 1: 0mn - no rewind
Note 2: fsf = move tape forward one record.

This is to get past the bootlif tape record.

# tar -tvf /dev/rmt/0m > /tmp/toc_file 2>&1

Note 3: 0m = for rewind. This will keep you on the tar record.

Note 4: get the table of contents or tape index.

# tar -xvf /dev/rmt/0m "./dir/file/*"

[Rajeev Shukla · Accepted solution]

The data on the other VG will not be effected as long as you choose right disks for VG00, Why i mean is, the disk for VG00 you choose should not clash with disks in other VG.
what i recomend is take a printout of other VG and lvmtab before you do restore, that will help you restore the other VG if they didn't exist when you took you'r last make_recovery.
strings /etc/lvmtab|lp -d
vgdisplay -v |lp -d

[Rajeev Shukla]

you could use the option given by Michael if you dont want any changes in the filesystem layout and just want restore all files.
But booting from tape will give you flexibility to change the filesystem sizes etc..

[Michael Steele_2]

Regarding your 2nd question: "...will my data on another vg be affected?"

Usually no. But it depends upon how the make_recovery tape was created. You can refer to the ignite logs or use the /tmp/toc_file from the above procedure to see what's on the tape.

Also, for an automatic restore: The "..no fuss..." procedure. Just boot off of the tape and leave the tape alone and don't interact with the keyboard.

However, mirroring will not be included in the restore. So bear this in mind especially with vg00, your pri and alt boot disks.

[Michael Steele_2]

Rajeev's procedure is the "Ignite Cloning" procedure and it involve's about 2 hours of work and should only be attempted after logical volume sizes, disk sizes, swap sizes, ip, hostname and all of the other network information has been collected. So there's some preparation to it.

The reason why its call cloning is because other HP servers of the same HW class can be installed with HP-UX this way.

This procedure is also found under 'Resizing Root, Swap and Boot Logical Volumes': If interested.

[Michael Steele_2]

I just realized, are you using "file system" as "operating system"?

Please clarify.

[Nufimi Zainal]

Mm,

Perhaps I have to explain my situation. Some error was made when a user tried to reduce the /opt directory. So right now, when I do a bdf:

Filesystem kbytes used avail %used Mounted on
/dev/vg00/lvol3 524288 86016 434880 17% /
/dev/vg00/lvol1 505392 58040 396808 13% /stand
/dev/vg00/lvol8 3080192 455744 2604696 15% /var
/dev/vg00/lvol7 2097152 1088320 1000976 52% /usr
/dev/vg00/lvol6 524288 86232 435336 17% /tmp
/dev/vg00/lvol13 4096000 2104 3838035 0% /oradata4
/dev/vg00/lvol12 4096000 2053962 1914414 52% /oradata3
/dev/vg01/lvol11 4096000 1999957 1965044 50% /oradata2
/dev/vg01/lvol10 4096000 2204745 1773057 55% /oradata1
/dev/vg01/lvol1 20480000 4898404 15338144 24% /oracle
/dev/vg00/lvol4 131072 2616 127512 2% /home
/dev/vg01/lvol2 20480000 93940 20067540 0% /appl

So the /opt directory is missing and fsck wouldnt work. So I have this make_recovery tape which I thought of using to recovery the /opt directory. I am not sure if the steps that I do is right. So hope you can assist me. Thanks

[Michael Steele_2]

Because you will be changing the state of the server you need to factor in the age of the make recovery tape and how stable the O/S has been since the tape was made.

The 'no fuss' automatic, pop the tape in and forget about it method is great as long as you remember to re-mirror everything, but it also depends on above factors.

Otherwise, because you're changing so much, especially if the tape is old, then I'd stick to the 'tar' procedure I listed.

[Michael Steele_2]

To verify the O/S:

# check_patches

# swverify -x=check_permissions=true \*

Check verify.log.

[Pratyush Paul_1]

Hi -


This is what I do in the data center.

#make_recovery -A ( complete vg00)

ensure I have a successful backup.

Now if you have to restore your system with this tape at any point, here is the option.

# interact with the boot process.

<> sea
ISL > bo p3 ( meaning p3 is my tape drive)

Do not do anything, just stay away from the system, it will go to the original root disk by default, it will link all you VGs as it was before. It will bring the system back with all the filesystem mounted.

Thanks

Pratyush