Hi,
I'm not very familiar with memory issues let alone OpenBSD's MFS memory
allocation/release algorithms, but the following strikes me as being a
bit odd:
net5501:
Filesystem 1K-blocks Used Avail Capacity Mounted on
/dev/wd0a 983486 286662 647650 31% /
mfs:26509 7855 2913 4550 39% /var
net4801:
Filesystem 1K-blocks Used Avail Capacity Mounted on
/dev/wd0a 983486 287002 647310 31% /
mfs:18470 7855 3271 4192 44% /var
net4521:
Filesystem 1K-blocks Used Avail Capacity Mounted on
/dev/wd0a 496894 424180 47870 90% /
mfs:22103 7855 5485 1978 73% /var
All boxes have an MFS partition mounted onto /var. Amount of RAM varies
from (by heart - I'm not sitting next to these boxes right now) 128MB
(net4521) to 512MB (net 5501). The 4521 has the least amount of memory,
that's for sure.
What happens is that the disk usage on the MFS partition stabilizes at
around 40% when I use a log rotating script. But with the same script,
the MU keeps growing on the 4521 towards 100%. A significant part of it
does not seem to be accounted for as actual disk usage:
df -h:
Filesystem 1K-blocks Used Avail Capacity Mounted on
mfs:22103 7.7M 5.4M 1.9M 74% /var
du -sh /var:
1.8M /var/
After a reboot it's back to a sane value (~20M). Leaks? Allocation
issues? As said, the net4521 is lowest on RAM.
Or am I staring the Stupid and Obvious right in the eye here?
Bill
_______________________________________________
Soekris-tech mailing list
Soekris-tech@...
http://lists.soekris.com/mailman/listinfo/soekris-tech
