|
Okay, I've run:
echo "4096" > /proc/sys/net/ipv4/neigh/default/gc_thresh3
then
ip route flush cache
The 'ip route flush cache' would need to be run only by the cron if we
don't fix the problem otherwise.
dmesg doesn't seem to be changing, so maybe that fixed it.
What do you mean? Can you repeat your ping test and tell me if you still
get the 'Neighbour table overflows', please?
And could you run the following for me please?
for i in /proc/sys/net/ipv4/neigh/*/gc_*; do
printf "%-48s ---> %s\n" "${i}" "$(cat ${i})";
done
inode_cache 114087 114149 512 16307 16307 1 : 124 62
dentry_cache 116490 116490 128 3883 3883 1 : 252 126
This has nothing to do with your problem, I was just surprised to see
this high number. This you could probably push down by changing the the
fs_commit rate and the bdflush intervall but this is yet another
problem. We solve your first problem first :)
Doesn't seem to have changed much. I'm not really sure what that all means,
but it doesn't seem to be getting bigger either.
If it works I will explain it unless Julian beats me to it.
Best regards,
Roberto Nibali, ratz
--
echo '[q]sa[ln0=aln256%Pln256/snlbx]sb3135071790101768542287578439snlbxq'|dc
| 