Hey,

i think i have a problem on a machine with trashing and in the follow
with the OOM Killer.
At 00:20 we make a backup from a Informix Database on disk.
In this time kills someone or somewhat our application server on the
same machine.
The application server is behind the Informix database the process
with the most forks, most ramuse, most cpu time.

Here some parameter form the machine at 00:00, 00:30, 00:31 and 00:40.
The application server crashs between 00:30 and 00:31.
My questions:
Is this trashing? When you need more or other informations lease post
this here.
How can i prevent processes from killing? echo -17 > /proc//maps/
oom_adj ?
Or can i tune the vm with the change of /proc/sys/vm/overcommit_ratio
and overcommit_memory?

Kernel: 2.6.5-7.244-bigsmp
RAM Total: 4GB
CPU: 4 Intel(R) Xeon(TM) CPU 2.80GHz

Parameter time time time time
00:00 00:30 00:31 00:40
/proc/sys/fs
dentry-state 117812-108623 5462-924 5377-839 7363-2808
inode-state 93469-8830 7198-1781 7177-1859 8266-1872

/proc/sys/vm
nr_pdflushthread 2 3 4 2

/proc/meminfo
memfree 945372 93836 92860 121420
cached 1695644 2793316 2794244 2776172
active 1726616 1556169 1555524 1563088
inactive 500536 1578428 1580104 1543280
lowfree 524268 92720 91496 83600
highfree 421104 1116 1364 37820
dirty 268 609764 341248 520
slab 108588 52124 51900 51456
commited_as 2347244 2347152 2346564 2324696

vmstat
free 945372 93836 93224 121296
inactive 500536 1578396 1579376 1543280
dentry-cache 118134-228144 5732-21Kernel190 5647-21060 7675-20956
buffer_head 132612-319713 402901-403919 403291-403990 387944-404416

Thanks

Sebastian