make_tape_recovey is failing

[Joe_419]

HP-UX 11.00 on a K480 with current patches installed.

I am trying to create an Ignite Backup tape and have done this many many times, but this time it is not completing. It gets to the writing boot lif to tape and then this happens:


* Writing boot LIF to tape
43470+1 records in
43470+1 records out
* Complete writing boot LIF onto tape.
* /opt/ignite/data/scripts/make_sys_image -c n -d /dev/rmt/1mn -t n -s
local -n 2005-06-23,17:18 -m t -w
/var/opt/ignite/recovery/2005-06-23,17:18/recovery.log -u -R -g
/var/opt/ignite/recovery/2005-06-23,17:18/flist -a 3284270

* Preparing to create a system archive

* Archiving contents of fast01 via tar image to local device/dev/rmt/1mn.
am24jun.txt : No such file or directory
am24old.txt : No such file or directory
am24jun.txt : No such file or directory
am24old.txt : No such file or directory
amjun05.txt : No such file or directory
amout.txt : No such file or directory
WARNING: The pax command returned a non-zero exit status (exit status 1)
* Creation of system archive complete
NOTE: The following files are in the list of files that were on the
system, but they are no longer present. These files are not
included in the back-up:
/var/tmp/ign_configure/make_sys_image.log

ERROR: The make_sys_image command failed. The system recovery archive will
not be created.

======= 06/23/05 17:57:37 MEST make_tape_recovery completed unsuccessfully

Has anyone ran across this. Any suggestions?

[Mel Burslan]

I am under the impression that you have run the make_tape_recovery command with the -x inc_entire=vg00 option.

And these am24* files were preset at the time ignite utility scanned all your vg00 for files and recorded them, but between the scan and the time it tried to write them to the tape, someone/some process removed these files.

This is the most likely scenario. If that is the case, and nobody is doing file deletions on vg00 mounted volumes anymore, you should not encounter this problem, if you run it one more time. But considering the nature of /tmp and /var/tmp, there will always be write's and delete's outside your control and more than likely they are mounted on vg00 unfortunately. So, short of taking the system offline to make a tape recovery archive, you can add the following option to your command line:

-x exclude=/tmp -x exclude=/var/tmp

also, it is advisable to add more excludes if you have any application or user who is doing file operations nortoriously over a vg00 filesystem. An mp3 downloader to his/her home directory would be a good candidate as an example.

Hope this helps

[Devender Khatana]

Hi,

The error indicated that the files were present when backup was fired but when it reached that file it was missing. The exit code has changed probably these files includes a ignite log filemake_sys_image.log which was removed from /var/tmp/ign_configure.

Try to isolate the cron /batch jobs which could have been doing this. Or could be space problem.

HTH,
Devender

[V. Nyga]

Hi Joe,

please give us your used command

Volkmar

[Alex Lavrov.]

I don't think that this is the reason for the failure ...

These are probably temp files that were present during the file list creation, but were removed before pax actually got to them and it's a normal condition. I have it almost every backup.

Do you see some more info in:
var/tmp/ign_configure/make_sys_image.log?

[Joe_419]

My command was as follows:

./make_tape_recovery -I -v -x inc_entie=vg00 -a /dev/rmt/1mn

Right now I am trying to run it by adding -x excludes for /tmp and for /var tmp so that is still running:

./make_tape_recovery -I -v -x exclude=/tmp -x exclude=/var/tmp -x inc_entire=vg00 -a /dev/rmt/1mn. It has not errored out yet but I am assuming it will and am not sure why. I will post again if it does error out.

[Joe_419]

I ran the command:

./make_tape_recovery -I -v -x exclude=/tmp -x exclude=/var/tmp -x inc_entire=vg00 -a /dev/rmt/1mn

It still failed, what gives?

Here is the output:

* Writing boot LIF to tape
43469+1 records in
43469+1 records out
* Complete writing boot LIF onto tape.
* /opt/ignite/data/scripts/make_sys_image -c n -d /dev/rmt/1mn -t n -s
local -n 2005-06-23,18:53 -m t -w
/var/opt/ignite/recovery/2005-06-23,18:53/recovery.log -u -R -g
/var/opt/ignite/recovery/2005-06-23,18:53/flist -a 3172160

* Preparing to create a system archive

* Archiving contents of fast01 via tar image to local device/dev/rmt/1mn.
am24jun.txt : No such file or directory
am24old.txt : No such file or directory
am24jun.txt : No such file or directory
am24old.txt : No such file or directory
amjun05.txt : No such file or directory
amout.txt : No such file or directory
WARNING: The pax command returned a non-zero exit status (exit status 1)
* Creation of system archive complete

ERROR: The make_sys_image command failed. The system recovery archive will
not be created.

======= 06/23/05 19:31:14 MEST make_tape_recovery completed unsuccessfully
#

[Joe_419]

Alex no make_sys_image.log in that directory. I searched for it and did not find it.

[Devender Khatana]

Hi,

See if you can write to that directory & there are no permissions / space issues.

HTH,
Devender

[Joe_419]

Yes I can rwx to that directory.

[Joe_419]

Is there an option to redirect where the tmp file gets put at. by default it goes to /tmp I believe or maybe /var/tmp. can I use some option to direct i under another directory. Although I do not believe this is the problem I was going to try it.

[Mel Burslan]

so. let me understand this correctly, you have run the make_tape_recovery command and it again failed due to not finding the very same files ?

Maybe you need to look into where these files are getting created and then removed from. It may not be under /tmp or /var/tmp. After finding it, you may need to exclude that directory or the files (with their full paths) by additional excludes.

[Devender Khatana]

Hi,

It seems the recovery is complete successfully as the recovery complete message is there. But the reason for error is the exit status which is because the log file has got removed where the final messages are to be put.

Can you try testing the ignite tape wheather it contains all files or not to confirm the validity of the same?

#export TMPDIR=/xyz
should change your /tmp location. But not sure wheather ignite will take it from here as it writes everything to /var/tmp.

Was it the first time the ignite was tried or was it working fine earlier ? Any recent changes ? Can you try to search for the missing file initially after starting backup ?

HTH,
Devender

[V. Nyga]

Hi,

I think about a wrong or not readable filename.
Try to find these files with:
'find / -name "*24old.txt*" -print'
Look if there's anything incorrect in the file name/path like '~' or 'space'.

I know this from a process I ran every week at my server.
UX is collecting the directories and purge this dirs. If there's a space in it (most common: 'Neuer Ordner') then UX can't look into this directory and fails.

HTH
Volkmar

[Cheryl Griffin]

Files that have unseen characters can cause these messages. To check if the filenames might be the problem:
# cd /var/tmp ; ll |vis

If am24jun.txt, am24old.txt, am24jun.txt, am24old.txt, amjun05.txt, and amout.txt are in /var/tmp - consider clearing /var/tmp. This is a directory that is for temporary application files ONLY (like vi etc.), and can be periodically cleared.

Check into configuring clean_tmps on reboot. This will clear /tmp and /var/tmp automatically for you at reboot. In the Knowledge Base look at "KBRC00012005: How can the clean_tmps start script remove files in /var/tmp?"

***** Note:
Documents in the Knowledge Base can be accessed if you have a valid support agreement and are entitled to access.

Posting or delivery of entitled material via the forums or email is in violation of the terms of usage. If you need help accessing these documents, contact HP Support.

[Doug_85]

Joe,

I just had the same error come up 2 days ago. We deleted the core dump files and used a new tape and the ignite tape was successfully completed.

Regards,
Doug