Amanda 2.4.2p2
Solaris 8
SDLT 220

We recently make a couple of changes to our amanda conf (well, one
of our amanda server configs).

1) changed the config name
2) changed from SW to HW compression, increased tape length
   in config file by 50 %, we where seeing nearly that in SW
   compression. The large partitions (just 2 of them) are Lotus
   Notes databases, they seem to compress well.
3) Moved from raid 0+1, um, mirrors that had been expended by
   concatination, on smaller drives 36 Gig range to RAID 5 on
   a new disk controller.

We are seeing that amanda used to perform level 0 and 1 dumps (DUMP
type dumps) and we are now seeing that we are getting to level 2
dumps. We where currious as to why.

Below are some of my musings, sent to the other admins at my site.

I'm really seeing this as a non-problem, however something is
acting "differently".

                                                thanks,

                                                Brian

----- Forwarded message from Brian Cuttler <brian AT wadsworth DOT org> -----

On Fri, Jan 21, 2005 at 10:35:36AM -0500, Quei-Len Lee wrote:
> 
> Something needs to be fixed that for several days, I see
> level 1 or 2 backups on the db directories, which is unusual.


Just thinking out loud... this will have to end up in my amanda-howto
(for the other local admins)

There are a couple of questions at the end.

It looks like there are a couple of things in play here.

1) From amanda.conf
bumpsize  50 MB         # minimum savings (threshold) to bump level 1 -> 2
bumpdays     1          # minimum days at each level
bumpmult     2          # threshold = bumpsize * (level-1)**bumpmult

This means that at least 1 days has to pass before the dump level
can be incremented. Also the dump level will not be incremented 
unless at least 50 MBytes in savings will be realized, that is from
level 0 to 1, from 1 to 2 you need to save 100 MBytes, from 2 to 3
150 MBytes. Amanda makes it progressively more difficult to reach
the bump threshold.

2) We have successful backups each night, even with the dump level
   being advanced we are backing up any files that are changed from
   the previous day. We have all changed files even, the intermediate
   versions of the databases, which we are seldom if ever asked to restore.

3) I had changed the tape type definition to reflect the length change
   representing HW compression being enabled. We are not incrementing
   to avoid EOT.

4) We did also change the crontab for bin to force level 0 dumps on
   Fridays. This change reflecting the changes made when we renamed
   "notes_dlt" to "wcnotes".

All of this looks ok to me. I do not know why amanda is incrementing
the dump levels more frequently than before but we are not losing any
data nor do I believe that a file restore (of individual files) would
be any more difficult than previous. A full restore would of course
entail restores from each dump level.

     Could we be seeing more change in the file system than previously
     because of changes to the disk cluster-size ? ie, the granularity
     in the number of blocks/sectors on the Raid vs the older single
     spindle disk drives ?

     We could increase both the bumpdays and bumpsize values to reduce
     the occurences of level 2, or increase the bumpmult value.

     We could even force level 0 every day if desired, backups seem to 
     complete quickly enough to allow level 0 every day without interfering
     with production.

I will ping the amanda-users list to see if they have seen anything similar.

---
   Brian R Cuttler                 brian.cuttler AT wadsworth DOT org
   Computer Systems Support        (v) 518 486-1697
   Wadsworth Center                (f) 518 473-6384
   NYS Department of Health        Help Desk 518 473-0773