Hello...
I setup amanda this past week. I had several questions, all of which
were answered by looking at the list archives. Thanks to all. Amanda
RULES!
I did run into a bug though. While I was doing test backups this is what
happened:
All scsi drives
DLT-8000 tape drive
using dump ( not tar)
Redhat linux (6.2)
mt-st-0.6
dump-0.4b20
amanda-2.4.2
mtx-1.2.11-pre1
holdingdisk hd1 {
comment "main holding disk"
directory "/var/spool/amanda"
use -512 Mb # how much space can we use on it
# a negative value mean:
# use all space except that value
chunksize 512Mb # size of chunk if you want big dump to be
This disk is a 18 gig scsi drive. At the time of this test ~8 gigs was
used elsewhere. So I had ~10 gigs free for amanda usage (10.5 gigs -
0.5 from holdingdisk). My disklist looked something like:
server1 /boot my-type
server1 / my-type
server1 /large my-type
...
server9 / my-type
server9 /var/spool my-type
All the small partitions went first and taped fine. The rest of the
larger partitions taped fine. Then the last two. Two ~20 gig
partitions with ~10 gigs used. The amstatus showed an estimate size of
9.9 gigs for boths these partitions. (The linux dump has a bug in the
size extimate). All the previous sessions ended up showing usage of
102% - 120%. The last two backups would fail and start over,
ping-ponging back and forth. It went something like this:
amanda-
I am backing up server3 /large with size estimate of 9.9gigs
I have 10gigs free on the holding disk -> save to disk
0%..10%...(backing up to disk)...100%...110%...
out of usable disk space -->> fail
Go to next disklist entry
I am backing up server4 /large with size estimate of 9.9gigs
I have 10gigs free on the holding disk -> save to disk
0%..10%...(backing up to disk)...100%...110%...
out of usable disk space -->> fail
Go to next disklist entry
( back to server6 and loop forever)
I saw this happen so I freed up more space on the holding disk. And it
kept happening even though 15 gigs were now free. The last two sessions
ping-ponged back and forth failing each time.
Maybe something like this could be added:
psuedo code
if (Out of usable disk space) {
happend before && issue warning && backup straight to tape
re-check disk to see if there is more space available
yes && continue
} else {
mark this session as needing ESTIMATE * 1.2 free disk space
or ESTIMATE = CURRENTLY_USED * 1.2
issue warning
try again
}
I think the whole problem comes from using the estimated size as a hard
number. And the current linux dump ( from sourceforge ), at least on my
servers, under reports disk usage. I am backing up 28 partitions and
every one of them shows 102% - 120% backed up.
Thanks to the developers. Amanda Rules!
--
Christopher McCrory
"The guy that keeps the servers running"
[EMAIL PROTECTED]
http://www.pricegrabber.com
"Linux: Because rebooting is for adding new hardware"
