memory leaks in 11.0?

[Konstantin Belousov]

On Wed, Jul 12, 2017 at 10:16:25AM +0200, Kajetan Staszkiewicz wrote:
> Dnia wtorek, 11 lipca 2017 17:12:05 CEST Konstantin Belousov pisze:
> > On Tue, Jul 11, 2017 at 03:56:04PM +0200, Kajetan Staszkiewicz wrote:
> > > `vmstat -z` shows constantly increasing allocation of "512" and "UMA
> > > Slabs". Memory allocated for all pf-related things seems fine. I have
> > > graphite graps for every `vmstat -z` and the icrease on "512" grows
> > > in similar way as "wired" memory. "512" has 2 917 392 used objects
> > > allocated at this moment, "UMA Slabs" is 379 006, there is 2636MiB
> > > "wired" memory.
> > 
> > UMA zone 512 is used for kernel mallocs of size less than 512 (and greater
> > than 256).  You can see mallocs types usage with vmstat -m.
> 
> Thank you for pointing me to that command, now I see that the biggest 
> allocation is:
> lltable 3240611 1620278K       -  3311016  256,512
> 
> I'll push all `vmstat -m` to graphite to see which one is growing.

lltable is the arp table, AFAIU.  Do you have something unusual configured
there, or some non-usual network activity ?

I have no idea if some leaks there were known and fixed since 11.0.