On Jan 07 15:14:21, [email protected] wrote:
> On Jan 07 13:27:35, [email protected] wrote:
> > This is current/amd64. Unlike in previous versions,
> > some memory-heavy processes (such as in building some
> > big port) get killed with
> >
> > UVM: pid 11422 (cc1plus), uid 0 killed: out of swap
> >
> > On this machine, I have _no_ swap. However, the machine has
> > 1G RAM, and most of it is free in the moment this happens.
> >
> > In fact, occasionaly, spawning a new xterm fails like this,
> > even if there is nothing else happening and there is nearly
> > 1G of free RAM.
> >
> > Has something changed in this respect?
> > Do I _have_ to have swap?
>
> On Jan 07 13:34:10, [email protected] wrote:
> > > UVM: pid 17950 (cc1plus), uid 0 killed: out of swap
> > > UVM: pid 11442 (cc1plus), uid 0 killed: out of swap
> > cc1plus is known to require vast ammounts of memory in some cases. It
> > is likely you *are* running out of swap.
>
> On Jan 07 14:03:08, [email protected] wrote:
> > 1G of ram is definitely not enough to build large ports on amd64.
> > Everything mozilla-related, for instance, goes up to 2G and more
> > while linking.
>
> This was indeed a build of gtk+3.
>
> On Jan 07 07:59:46, [email protected] wrote:
> > I ran into /etc/login.conf limits of datasize = 512M way before
> > hitting any other limit, so is that bumped?
>
> This was running as root, who is staff:
No, sorry; root is in the daemon class:
daemon:\
:ignorenologin:\
:datasize=infinity:\
:maxproc=infinity:\
:openfiles-cur=128:\
:stacksize-cur=8M:\
:localcipher=blowfish,8:\
:tc=default:
