Errors about make_recovery

[LUO Lin_2]

Hi, all there

I made a bakeup with
# make_recovery -A
then boot from the backuped tape. After Creating logical volume ... & Extanding logical ..., got some errors as follows:
==============================================

Error: source_read, read failed: I/O error (errno=5)
Error: sourcefile_open: source read: I/O error (error=5)
Error: Tar: blocksize=0; broken pipe?
Error: File: /sbin/fs/hfs/mkfs not found
Error: File: /sbin/fs/hfs/newfs notfound

========================================
OS: HP_UX 11.0
H/W: R380

I use a external Tape connected to SE SCSI2 port, which has the same H/W path with root disk. Is this a problem?

Thks in advance.

[Jim Mulshine]

Perhaps the make_recovery backup is corrupt. To verify this, if you have an older make_recovery from the same system you can try booting on that tape. Or, if you have another system that can be overwritten you could test the troublesome make_recovery tape on that system.

[Pedro Sousa]

Hi!
I don't understand how you did make the tape!!
It cannot be on the same path! Change the scsi ID, and make a new tape.
good luck.

[LUO Lin_2]

Thks Jim.

First, how to test the troublesome tape?

Second, no any errors displayed when "make_recovery -A" finished. So I doubt there is something wrong with the old tape.

Any idea else?

[LUO Lin_2]

Thks Pedro. First make clear something: I meant the tape & root disk are on the same scsi bus such as 8/16/5, not same scsi target id.

Looking forward to your further tips.

[Pedro Sousa]

OK!
Check your tape with "check_recovery".

[Vincenzo Restuccia]

Try with:
#mt -t /dev/rmt/omn fsr 1 (skip boot area)
#tar tv

[Bill McNAMARA_1]

verify your tape device file
ioscan -fnkCtape

make_recovery -Av -d /dev/rmt/0m

check_recovery as mentioned to verify.

Also make sure that you recover to a similar system, same config/hardware.

Get the latest ignite software
www.software.hp.com/products/IUX/

Get the latest PAX patch for your OS version,
search here on the itrc.

Verify your /etc/fstab before creating the tape. mount -a, and watch out for networked filesystems when using the -A option, especially /home

Later,
Bill

[Patrick Wallek]

check_recovery will NOT verify that the make_recovery tape that you made is valid. check_recovery only works if you use the '-C' option to make_recovery and then when check_recovery is run it compares your system to a status file to tell you if you need to run another make_recovery.

Excerpt from check_recovery man page:

check_recovery - compare the current system to the System Recovery status file created by the last invocation of make_recovery.

check_recovery compares the current state of the system to the System Recovery status file (created by the last invocation of make_recovery) to determine if a new System Recovery Tape needs to be created. Only Core OS and User Core OS files are validated. Refer to make_recovery for an explanation of these terms.

To check your tape I would do what was recommended in a previous response:

mt -t /dev/rmt/?mn fsf 1 --- (10.20 syntax)
mt -f /dev/rmt/?mn fsf 1 --- (11.X syntax)

tar -tvf /dev/rmt/?mn --- let this run and see if it errors. If it does you have a bad tape.

If the tape is bad I would do 2 things: 1) Clean the tape drive thoroughly and 2) use a new tape and make another make_recovery, if possible.

When doing the make_recovery I always use the "no rewind" device file. The man page for make_recovery does say that this is required. I don't know if it will make a difference or not, but give it a try.

make_recovery -Avd /dev/rmt/?mn is what I would recommend.