On Fri, Sep 23, 2005 at 07:28:33AM -0400, Bernard_Wu AT NCCI DOT COM wrote:
>
>
> Hi Listserve,
> I am trying to configure Amanda to run incrementals only. The full backups
> are done outside Amanda. Last night , I ran my first test. I am trying to
> backup just /etc. Here are the results.
It is good your are trying just one small item. Once that works you
can always expand the disklist.
You don't really have to wait for the evening each test. amdump can be
run from the command line using the same command used in your crontab.
To be more precise, you should be running the amanda user, using /bin/sh
and have a minimal environment, including the minimal PATH. Make a copy
of your config directory before a run from the command line. Before each
new test run, replace each of the changed files with copies of the original.
>
> These dumps were to tape dump200.
> The next tape Amanda expects to use is: a new tape.
> The next new tape already labelled is: dump300.
>
> FAILURE AND STRANGE DUMP SUMMARY:
> lnxdev.ncc /etc lev 0 FAILED [dump larger than tape, -1 KB, but cannot
> incremental dump new disk]
> planner: FATAL cannot fit anything on tape, bailing out
>
>
> STATISTICS:
> Total Full Daily
> -------- -------- --------
...
>
>
> NOTES:
> driver: WARNING: /amholdsk: not 4398046511104 KB free.
> planner: tapecycle (5) <= runspercycle (20)
fix this
...
>
> My questions follow;
>
> 1. Why does Amanda think that /etc is larger than 1 tape, or is it
> because it does not have a level 0 dump to compare against ?
That is what I was talking about in my response to your first question.
What will amanda increment against??
The "-1 KB" in some parts of your report is just an impossible size in
the real world used as an error value for some function returns (like
estimated size of a level 0, a level 1, ...)
> 2. Do I have to force a level 0 dump of /etc to jumpstart the process ?
That is one way, but as I previously noted, all your subsequent incrementals
will be based on this initial level 0. The won't be reset to the date of
the level 0 done by your system. So they will get larger and larger.
I think amanda used to use a file /etc/amandates to track when levels were
last done. It may have switched to following log files info rather than
amandates, I don't recall. But if amandates is still used you may be able
to trick it into thinking a level 0 was done. If it works, set up a mechanism
whereby the file gets adjusted each time the system does a level 0. If a
switch to log files has been made a similar approach might be possible.
With no level 0 available, I don't know what amrecover will do. amrestore
should be fine.
Here is a portion of my /etc/amandates file.
/ 0 1104304406
/ 1 1104821711
/ 2 1096005295
/ 3 1062223755
//tec/C 0 1099551707
//tec/C 1 1100070105
//tec/D 0 1099724570
//tec/D 1 1100069872
First column is the host/disk name.
Second is the level number.
The final column is a time stamp in "unix time",
i.e. seconds since 1970 using the GMT (aka UTC) timezone.
If your date command has gnu-type options it can give
unix time for an arbitrary date/time. For example,
on Solaris the gnu version of date is gdate. So assuming
your system level 0 was completed on Sep 21 at 2:22 AM and
44 seconds you could do something like this:
gdate -s '09/21/2005 02:22:44' +%s
1127283764
Then your /etc/amandates file could be created with:
/etc 0 1127283764
And do your test. Otherwise what logfile info amanda
uses will have to be tracked down and a similar approach
developed for that.
--
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)
|
