LinuxQuestions.org - Questions about Tape Backup for Sys Admins

- Linux - Newbie (https://www.linuxquestions.org/questions/linux-newbie-8/)

- - Questions about Tape Backup for Sys Admins (https://www.linuxquestions.org/questions/linux-newbie-8/questions-about-tape-backup-for-sys-admins-662799/)

Here are the details of the 4 modes available:

mode1 blocksize=0 density=0x51 compression=1 # DLT-V4 density, compression on
mode2 blocksize=0 density=0x51 compression=0 # DLT-V4 density, compression off
mode3 blocksize=0 density=0x50 compression=1 # VS160 density, compression on
mode4 blocksize=0 density=0x40 compression=1 # VS80 density, compression on

The unit is a Quantum DLT-V4 and the tapes are HP DLT Tape VS1 (160 GB with 2:1 compression).

I guess I should use mode 1 or mode 3. Not sure.

When I tar the data, should I also zip or rely on the tape compression?

OK, mode 1 is what I need for sure, or mode 2 if I want to disable tape compression.

I went ahead and started the backup *crosses fingers*

Will users notice a serious drop in performance as the tar program reads every single file on the server? CPU usage looks minor, but I'm worried about disk access (I guess that's a good reason to do it during low usage time).

I decided I'm not worried if a few files get corrupted. I think it's reasonably likely that none will anyway. Each file (unless it happens to be very big) would be written to the tape very quickly, so it's not *too* likely that the file will be in the process of being modified when the tar program picks it up (unless the file is modified very frequently).

I think I'm getting the hang of this. The big server needs two tapes. I think the data will fit on the tapes using the configuration below. I test compressed a 100 MB of mail and it was neatly cut in half so I know it will all fit. Same with the second tape.

Any reason the following scripts wouldn't work? Any suggestions to improve them?

Code:

#!/bin/csh



mt -f /dev/nsa0 rewind



tar cvzf /dev/nsa0 /dev      #individual partitions

tar cvzf /dev/nsa0 /local

tar cvzf /dev/nsa0 /mail

tar cvzf /dev/nsa0 /scratch

tar cvzf /dev/nsa0 /soft

tar cvzf /dev/nsa0 /tmp

tar cvzf /dev/nsa0 /usr

tar cvzf /dev/nsa0 /var

tar cvzf /dev/nsa0 /u1



tar cvzf /dev/nsa0 /1* #all files and folders on root partition

tar cvzf /dev/nsa0 /2*

tar cvzf /dev/nsa0 /3*

tar cvzf /dev/nsa0 /5*

tar cvzf /dev/nsa0 /6*

tar cvzf /dev/nsa0 /7*

tar cvzf /dev/nsa0 /8*

tar cvzf /dev/nsa0 /A

tar cvzf /dev/nsa0 /C*

tar cvzf /dev/nsa0 /b*

tar cvzf /dev/nsa0 /c* #does it matter if I pick cdrom?

tar cvzf /dev/nsa0 /d* #does it matter if I pick up dev?

tar cvzf /dev/nsa0 /e*

tar cvzf /dev/nsa0 /h*

tar cvzf /dev/nsa0 /i*

tar cvzf /dev/nsa0 /k*

tar cvzf /dev/nsa0 /li*

tar cvzf /dev/nsa0 /lp*

tar cvzf /dev/nsa0 /mnt #does it matter if I pick up mnt?

tar cvzf /dev/nsa0 /mo*

tar cvzf /dev/nsa0 /n*

tar cvzf /dev/nsa0 /p*

tar cvzf /dev/nsa0 /r*

tar cvzf /dev/nsa0 /sbin

tar cvzf /dev/nsa0 /stand

tar cvzf /dev/nsa0 /sy*

tar cvzf /dev/nsa0 /tempmail

tar cvzf /dev/nsa0 /usbd.corer



mt -f /dev/nsa0 rewind

AND:

Code:

#!/bin/csh

mt -f /dev/nsa0 rewind

tar cvzf /dev/nsa0 /u2

tar cvzf /dev/nsa0 /u4

mt -f /dev/nsa0 rewind

OK, I ran a test script:

Code:

#!/bin/csh

mt -f /dev/nsa0 rewind

tar cvzf /dev/nsa0 /root/z/dir1

tar cvzf /dev/nsa0 /root/z/dir2

tar cvzf /dev/nsa0 /root/z/dir3

mt -f /dev/nsa0 rewind

Then I attempted to check the tables to verify that everything was alright, but I got errors:

Code:

srv# tar tvzf /dev/nsa0

gzip: stdin: decompression OK, trailing garbage ignored

drwxr-xr-x root/wheel        0 Aug 15 19:38 2008 root/z/dir1/

-rw-r--r-- root/wheel        0 Aug 15 19:38 2008 root/z/dir1/filez1

-rw-r--r-- root/wheel        0 Aug 15 19:38 2008 root/z/dir1/filze2

-rw-r--r-- root/wheel        0 Aug 15 19:38 2008 root/z/dir1/filze3

tar: Child returned status 2

tar: Error exit delayed from previous errors



srv# tar tvzf /dev/nsa0

drwxr-xr-x root/wheel        0 Aug 15 19:38 2008 root/z/dir2/

-rw-r--r-- root/wheel        0 Aug 15 19:38 2008 root/z/dir2/filez4



gzip: stdin: decompression OK, trailing garbage ignored

-rw-r--r-- root/wheel        0 Aug 15 19:38 2008 root/z/dir2/filez5

-rw-r--r-- root/wheel        0 Aug 15 19:38 2008 root/z/dir2/filez6

tar: Child returned status 2

tar: Error exit delayed from previous errors



srv# tar tvzf /dev/nsa0

drwxr-xr-x root/wheel        0 Aug 15 19:38 2008 root/z/dir3/



gzip: stdin: decompression OK, trailing garbage ignored

-rw-r--r-- root/wheel        0 Aug 15 19:38 2008 root/z/dir3/filez7

-rw-r--r-- root/wheel        0 Aug 15 19:38 2008 root/z/dir3/filez8

-rw-r--r-- root/wheel        0 Aug 15 19:38 2008 root/z/dir3/file9

tar: Child returned status 2

tar: Error exit delayed from previous errors

I added a space before each command, but the other spaces were in the output.

err..

Maybe running ...

tar cvf /dev/rst12 /

... on the other server wasn't such a good idea. This /proc/ directory is causing all kinds of trouble and after looking up info about it... it's probably a bad idea to include it.

Is there any way to tell the system to tar / but skip /proc ? If so, is there anything else I should skip? Is this going to ruin the backup (it's giving a lot of errors (arg list too long, is not a file, file changed size, invalid argument) although it keeps on going).

It's also seriously degrading system performance although no one is using this system but me at the moment thankfully.

EDIT:

Well, it made it through finally. I'm coming to work tomorrow to back up the other server if I can get some good intel from you guys. Thanks :) Otherwise... it might have to wait until monday night though things are going to get busier and busier ...

Yeah, don't backup /proc or /dev, they're really pseudo type filesystems, they don't really exist when you turn the machine off and created when the machine is turned on.

I also usually exclude /tmp and some of /var where I don't need to backup logs as I usually have a centralized loghost anyways.

And if you don't have a lot of customized applications in /usr, you can usually omit that as well since you can do bare metal recoveries by installing the base OS first, then applying the backed up data, etc.

Quote:

Originally Posted by trickykid (Post 3249621)

On one of the systems I'm using, the previous admin made a directory:

/var/adm/lists

And there are some important scripts in it. For now can't skip anything except /proc and /dev

Unless there is a really large chunk of data somewhere that I can be sure is not vital (as I could use a bit of space).