Re: mbuf cluster leak?

[Edgar Fuß <ef%math.uni-bonn.de@localhost>]

So this keeps happening and suddenly started without any configuration changes.
The only thing I changed (some two weeks before it happened first time) was 
to bump IPFilter's mabucket and logsize.

I wrote a collectd plugin to monitor mclpl and now have nice Grafana graphs.
What happens is that mclpl stays constant and then suddenly starts to rise
approxymately linearily (at a rate of around 4k per hour), suddenly stops 
rising and stays constant again.

This only happens on the gateway, not on several similar non-gw servers.
I tried ipfilter restart, but that doesn't help.

The weird thing is that vmstat -m shows ZERO mclpl Reclaims, but only on that 
one machine.

We now reboot the gateway every few days (as soon as mclpl has risen too much), 
but that's pretty awkward for a server that used to run non-stop for years.

Any hints? Any way to log mclpl Requests? That's only 16k in two days, so 
I guess wouldn't explode the logs if done.