(Partially) solved problems with lab10

[Noah Scales]

- using the original notice syntax *.=notice failed to give me messages, but *.notice worked, and then using the original syntax worked (changing syslog.conf, restarting syslog, and then checking notices file using tail -f), so it must have been my mistake in the first place.

- forgetting to escape the "%" in my date specification bombed my cron script, but looking at the slides would have helped. Doh.

- dump does not allow level1 backups of subdirectories (message in cron mail). According to the slides it does.

- dump does not let you update the dumpdates file when dumping a subdirectory (message in cron mail). According to the slides it does.

I haven't resolved those last two problems yet. Any advice?

-Noah

[Rich Simms]

What commands are you using in your cronjob?
What message are you getting in cron mail?

Here is a little example of doing full (level 0) and incremental (level 1) backups of the /opt directory and subdirectories. Between the backups I modified two files in the /opt/lampp/htdocs subdirectory. The level 1 dump got both of them with no problem.

[code][root@benji full]# cd /
[root@benji /]# dump 0uf /backup/full/opt-full-12-07-2008.dmp /opt
DUMP: Date of this level 0 dump: Sat Dec 6 03:54:29 2008
DUMP: Dumping /dev/sda5 (/opt) to /backup/full/opt-full-12-07-2008.dmp
DUMP: Label: /opt
DUMP: Writing 10 Kilobyte records
DUMP: mapping (Pass I) [regular files]
DUMP: mapping (Pass II) [directories]
DUMP: estimated 238165 blocks.
DUMP: Volume 1 started with block 1 at: Sat Dec 6 03:54:31 2008
DUMP: dumping (Pass III) [directories]
DUMP: dumping (Pass IV) [regular files]
DUMP: Closing /backup/full/opt-full-12-07-2008.dmp
DUMP: Volume 1 completed at: Sat Dec 6 03:55:33 2008
DUMP: Volume 1 266400 blocks (260.16MB)
DUMP: Volume 1 took 0:01:02
DUMP: Volume 1 transfer rate: 4296 kB/s
DUMP: 266400 blocks (260.16MB) on 1 volume(s)
DUMP: finished in 61 seconds, throughput 4367 kBytes/sec
DUMP: Date of this level 0 dump: Sat Dec 6 03:54:29 2008
DUMP: Date this dump completed: Sat Dec 6 03:55:33 2008
DUMP: Average transfer rate: 4296 kB/s
DUMP: DUMP IS DONE
[root@benji /]# cat /etc/dumpdates
/dev/sda5 0 Sat Dec 6 03:54:29 2008 -0800
[root@benji /]# cd /opt/lampp/htdocs/
[root@benji htdocs]# ls
favicon.ico index.html sample.html webalizer xampp
[root@benji htdocs]# vi index.html
[root@benji htdocs]# vi sample.html
[root@benji htdocs]# touch index.html
[root@benji htdocs]# cd /
[root@benji /]# dump 1uf /backup/incremental/opt-inc-12-07-2008.dmp /opt
DUMP: Date of this level 1 dump: Sat Dec 6 03:57:57 2008
DUMP: Date of last level 0 dump: Sat Dec 6 03:54:29 2008
DUMP: Dumping /dev/sda5 (/opt) to /backup/incremental/opt-inc-12-07-2008.dmp
DUMP: Label: /opt
DUMP: Writing 10 Kilobyte records
DUMP: mapping (Pass I) [regular files]
DUMP: mapping (Pass II) [directories]
DUMP: estimated 39 blocks.
DUMP: Volume 1 started with block 1 at: Sat Dec 6 03:58:02 2008
DUMP: dumping (Pass III) [directories]
DUMP: dumping (Pass IV) [regular files]
DUMP: Closing /backup/incremental/opt-inc-12-07-2008.dmp
DUMP: Volume 1 completed at: Sat Dec 6 03:58:02 2008
DUMP: Volume 1 50 blocks (0.05MB)
DUMP: 50 blocks (0.05MB) on 1 volume(s)
DUMP: finished in less than a second
DUMP: Date of this level 1 dump: Sat Dec 6 03:57:57 2008
DUMP: Date this dump completed: Sat Dec 6 03:58:02 2008
DUMP: Average transfer rate: 0 kB/s
DUMP: DUMP IS DONE
[root@benji /]# restore -tf /backup/incremental/opt-inc-12-07-2008.dmp
Dump date: Sat Dec 6 03:57:57 2008
Dumped from: Sat Dec 6 03:54:29 2008
Level 1 dump of /opt on benji.localdomain:/dev/sda5
Label: /opt
2 .
22089 ./lampp
32213 ./lampp/htdocs
32338 ./lampp/htdocs/index.html
32172 ./lampp/htdocs/sample.html
[root@benji /]# cat /etc/dumpdates
/dev/sda5 0 Sat Dec 6 03:54:29 2008 -0800
/dev/sda5 1 Sat Dec 6 03:57:57 2008 -0800
[root@benji /]#
[/code]
- Rich

[Noah Scales]

Oops, sorry, yes, actually, I (think I) used exactly what showed in your slides, but I'm not in front of a benji computer right now, so my response with actual code will have to wait. --Noah

[Rich Simms]

Note that Lab 10 is less complex than the Lesson slides. For the lab you are only doing Level 0 and 1 backups. In the Lesson we did level 0, 1 and 2 backups.

I also see a typo on step 8 which should read:

Schedule a level [color=#FF0000]1[/color] backup to run each minute for six minutes following the level 0 backup. Let me know what you find when you get back to a computer.

- Rich

[patrick weaver]

I have been working on Lab10 and Have run into the same difficulties mentioned by Noah. While I was able to execute the dump upon /opt using the syntax which you printed here, the same command when applied to the /home dir (per the asst.) fails with the following error messages:
"DUMP: You can't update the dumpdates file when dumping a subdirectory
DUMP: The ENTIRE dump is aborted."
The dump of /opt is reclassified as "Dumping /dev/sda5 (/opt)" I am wondering if this may be a clue. Is it possible that a /dev/* may only be dumped in its entirety?? I have been trying to think of a compelling reason for the policy which Noah and I have been stubbing our toes on. A look at the output of <mount> shows that there is no /dev/sd* = /home; does there need to be one? (the dump command works on /dev/sda3 (/var). A peek at lab09.pdf shows that /home is supposed to be mounted upon /dev/sda7 for the machine to qualify as purebred maltese. The central CTC benji does not meet these reqs as it has only six /dev/sda's. I am endeavouring to fix this now ( 12-08-08 2:20PM). -P

[Hilde Bell]

I'm running into the same problem. The e-malesDump does not want to update the dumpdates file and aborts the whole dump. Where is the dumpdates file? Is it trying to dump something that it's writing to? Is there a dumpdates inside /home?

[Joe Bobisuthi]

The dumpdates file is in /etc I believe, so that's unlikely to be the problem.

[Hilde Bell]

Is there a different file inside /home to which dump needs to write?
Also, I just noticed that I'm also getting the problem with notices now. The syslog.conf file worked fine for part one of the lab, but now it's messing something up in the cron jobs and in the dump command. I just tried to run the exact same command as in the example above, but was unable to. The dump command wants an existing file to dump /opt into.
I'm confused

[Rich Simms]

Oh ... I see the problem!

dump only works on complete filesystems. Unlike cpio and tar you cannot backup files or subdirectories. If you are using a pristine Benji VM then /home is still a sub-directory in /dev/sda1. If Lab 7 is applied, then /home is on its own partition and on a separate and complete filesystem.

Workaround for Lab 10: If you can't backup /home with dump then you can backup /opt. /opt in mounted on /dev/sda5 which has a complete ext3 filesystem.

Sorry for the confusion!

- Rich

[Hilde Bell]

Thanks, I'll try that.