amdump freezes

hello,

some clients with "big" partitions (>100Gbytes) freezes my amdump, Iusually get dumps errors which cannot end properly.

I have 2 questions,
1) how can I resolve that "client" error, timeout or whatever ?

2) why the amanda server doesnt giveup after dtimeout (1800s) andfinishes its amdump ?

I use amanda-2.5.0p2-4 on a centos 5.1 server with disk (virtual tapeson a raid 5) backup media.


here's the client error:

sendbackup: time 2185.293: 87: normal(|): DUMP: 4.10% done at 2642kB/s, finished in 13:38sendbackup: time 2485.300: 87: normal(|): DUMP: 4.67% done at 2632kB/s, finished in 13:37

....

sendbackup: time 33385.302: 87: normal(|): DUMP: 60.34% done at 2453kB/s, finished in 6:04sendbackup: time 33685.307: 87: normal(|): DUMP: 60.89% done at 2453kB/s, finished in 5:59

sendbackup: time 33753.599: index tee cannot write [Broken pipe]
sendbackup: time 33753.599: pid 25328 finish time Sun May 25 07:37:50 2008
sendbackup: time 33753.600: 109:  normal(|):
sendbackup: time 33753.601: 112: strange(?): gzip: stdout: Broken pipe

sendbackup: time 33753.601: 112: strange(?): sendbackup: index teecannot write [Broken pipe]

sendbackup: time 33753.610:  87:  normal(|):   DUMP: Broken pipe

sendbackup: time 33753.611: 87: normal(|): DUMP: The ENTIRE dump isaborted.sendbackup: time 33753.611: error [compress returned 1, /sbin/dumpreturned 3]

sendbackup: time 33753.611: pid 25325 finish time Sun May 25 07:37:50 2008


[root@backup /var/lib/amanda/int]
$ amstatus int --dumping
Using /var/lib/amanda/int/amdump.1 from sam mai 24 22:06:48 CEST 2008

helios:/home4 0 85174m dumping 37083m( 43.54%) (22:22:03)helios:/home9 0 109374m dumping 30169m( 27.58%) (22:15:15)



amanda.conf:

etimeout -1200
dtimeout 1800
tpchanger "chg-multi"

define dumptype bi-comp-user-size {
   comp-user
   comment "Non-root partitions on bi-proc machines"
   maxdumps 2
   estimate calcsize
}
Disklist exemple of big partition
helios /home4   bi-comp-user-size

Thanks .