On Sun, Aug 17, 2003 at 08:16:17AM -0300, Fernando Shayani wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> I'm having a problem with Amanda.
>
> I configured my Amanda to backup some FS in 1 tape... All filesystems
> toghether are 7GB and my tape is 12GB (SONY DDS-3).
>
> Well, I want to make incremental backups everyday, and a full backup
> after 7 days.
Putting on my nasty hat, in your previous message you asked for a
definition of an incremental backup, but you know you want them
every day?
> ... This is my amanda.conf file (just the relevant info).
Just what you think relevant. But leaving out some things I
think are probably relevant. If you want to leave things out
for space considerations, best to drop comments, unused tapetypes,
and unused dumptypes.
> - -----------------------
> inparallel 4
> netusage 600 Kbps
>
> dumpcycle 1 weeks
> runspercycle 0
This says you will run amdump 0 (zero) times each week.
I hope that is not true.
> tapecycle 1 tapes
This says you only have one tape to use.
I hope that is not true.
> bumpsize 20 Mb
> bumpdays 1
> bumpmult 4
With a tape larger than your data, these defaults can be left alone.
> etimeout 300
>
> runtapes 1
>
> tapedev "/dev/tape"
Is this a "NO-rewind on close" device? It must be!
Usually there is an "n" somewhere in the device name
for the no-rewind version of the device.
> tapetype SONY-DDS3
> labelstr "DIARIO"
This label string says every tape you have to use will be
labeled exactly "DIARIO". That doesn't let amanda distinguish
one tape from the next. I fear that you really do think you
can do backups with a single tape.
> holdingdisk hd1 {
> ~ comment "main holding disk"
> ~ directory "/var/tmp"
> ~ use 2000 Mb
> }
> - --------------------------------------------------------
>
> The problem is: I receive the report everyday saying the backup was
> made, but I cannot locate any file!!!
> I run amrecover, set the host, set the disk, I can see the directoryes,
> but they are all empty! I'm pasting the last REPORT here:
Do you have indexing turned on? (one of those missing "irrelevant infos")
What version of tar are you using. This is one possible symptom of a
buggy tar (tar --version). Try for either 1.13.25 or 1.13.19 (not 1.13).
>
> These dumps were to tape DIARIO.
> The next tape Amanda expects to use is: DIARIO.
Yup, looks like only one tape.
Honest, that is not a good idea.
Each run of amdump overwrites the previous run on the tape.
> STATISTICS:
> ~ Total Full Daily
> ~ -------- -------- --------
> Estimate Time (hrs:min) 0:00
> Run Time (hrs:min) 0:11
> Dump Time (hrs:min) 0:04 0:03 0:00
> Output Size (meg) 508.0 501.8 6.2
> Original Size (meg) 564.4 554.2 10.1
> Avg Compressed Size (%) 90.0 90.5 61.3 (level:#disks ...)
> Filesystems Dumped 13 2 11 (1:8 2:3)
As amanda was designed to do, on each run of amdump, it is giving you
a mixture of level 0 (aka full dumps) for some FS's (aka DLE's or
DiskListEntries), and some higher levels (eight level 1's and three
level 2's, aka incrementals) for the other DLE's.
With amanda's scheduling scheme, and your settings assuming runspercycle
is really 7, amanda will try to backup about 1/7th of your data each run
so it should average about 1GB/run. However, with so few DLE's, on a
day to day basis the amount of data backed up will bounce around somewhat.
So it looks like amanda is doing its thing correctly.
But I still fear that each run of amdump is trashing your previous run.
> Avg Dump Rate (k/s) 2424.4 2662.4 295.0
>
> Tape Time (hrs:min) 0:09 0:08 0:01
> Tape Size (meg) 508.7 501.9 6.8
> Tape Used (%) 4.4 4.3 0.1 (level:#disks ...)
> Filesystems Taped 13 2 11 (1:8 2:3)
> Avg Tp Write Rate (k/s) 978.4 1128.1 91.2
>
>
> NOTES:
> ~ planner: Incremental of sedenet:/SEDENET/psts bumped to level 2.
> ~ planner: Incremental of sedenet:/SEDENET/groups/secext bumped to level 2.
> ~ planner: Incremental of sedenet:/SEDENET/groups/secnac bumped to level 2.
There were a lot of changes on these DLE's so it was more efficient to
backup just the changes from the last level 1 rather than the last level 0.
Thus these backups were "bumped" to level 2.
> ~ planner: Full dump of sedenet:/SEDENET/fotos promoted from 4 days ahead.
> ~ planner: Full dump of sedenet:/home promoted from 4 days ahead.
In trying to balance the daily dumps to that 1GB/day I mentioned, amanda
decided to do a level 0 of these DLE's earlier that required by your setting
of a 1 week dumpcycle.
> ~ taper: tape DIARIO kb 520928 fm 13 [OK]
13 DLE's were sent to this tape (fm 13 == file marks).
The total data sent to the tape was 0.52 GB (post software compression).
> DUMP SUMMARY:
> ~ DUMPER STATS TAPER STATS
> HOSTNAME DISK L ORIG-KB OUT-KB COMP% MMM:SS KB/s MMM:SS KB/s
> - -------------------------- --------------------------------- ------------
> sedenet -ENET/fotos 0 351290 341728 97.3 2:052737.4 5:001140.3
> sedenet -oups/ascom 1 270 96 35.6 0:00 246.1 0:02 49.4
> sedenet -groups/cnp 1 140 64 45.7 0:00 207.4 0:14 4.7
> sedenet -roups/info 1 10 64 640.0 0:00 0.0 0:02 32.8
> sedenet -ups/secext 2 830 192 23.1 0:00 313.9 0:18 10.5
> sedenet -ups/secnac 2 810 224 27.7 0:01 239.1 0:02 92.3
> sedenet -ups/tesnac 1 260 96 36.9 0:00 97.1 0:12 7.8
> sedenet -DENET/psts 2 40 64 160.0 0:00 40.1 0:02 28.3
> sedenet /home 0 216220 172192 79.6 1:082525.0 2:361104.5
Check out the columnspec settings to adjust your column widths.
> sedenet -ysql/ponto 1 10 64 640.0 0:00 0.0 0:13 5.0
> sedenet -mysql/siri 1 10 64 640.0 0:00 0.0 0:02 27.3
> web.aen.baha /home/www 1 120 64 53.3 0:01 11.8 0:02 32.8
> web.aen.baha -spool/mail 1 7890 6016 76.2 0:19 311.4 0:07 855.6
>
> (brought to you by Amanda version 2.4.3)
> - ----------------
>
> Can anyone help me understand it? What happend???
HTH, generally looks good.
--
Jon H. LaBadie jon AT jgcomp DOT com
JG Computing
4455 Province Line Road (609) 252-0159
Princeton, NJ 08540-4322 (609) 683-7220 (fax)
|