Re: [solved] What am doing wrong

[Al <al%familysafeinternet.com@localhost>]

Hi,

After much testing on my test box, I realized that the problem was that I 
had a problem with the network card. It was giving me errors. I was first 
using a 3com card and then an Intel Pro 100. Both seemed to give me a lot 
of timeout errors. In fact, the intel card would stop working for few 
seconds each time. After replacing it with a realtek card the problem went 
away. This also stopped the issue of nginx and apache going into a state 
of sokva.

Thanks,
Al

On Mar 5, 2013, at 10:13 PM, Donald Lee wrote:
> I think a 16 MB sendbuf/rcvbuf may well cause timeouts if the network
> isn't pretty fast *and* reliable.  256K is usually plenty, which is the
> default.
>
> -dgl-
> >
> > I just did a fresh install of NetBSD 5.1. I then built nginx, php-fpm, 
> > and mysql. I then tested it buy running 10,000,000 dynamic php hits. Now, a 
> > couple of days later, some processes had a stat of sokva.
> >
> > I did build a custom kernel, but the only things that I changed were:
> > maxusers        64
> > options         NMBCLUSTERS=16384
> > config          netbsd  root on sd0a type ffs
> > #pseudo-device  raid            8       # RAIDframe disk driver
> > #options        RAID_AUTOCONFIG         # auto-configuration of RAID 
> > components
> >
> > And my sysctl.conf looks like this:
> > kern.no_sa_support?=1
> > ddb.onpanic?=0
> > kern.somaxkva=67108864
> > kern.maxfiles=16384
> > net.inet.tcp.recvbuf_auto=1
> > net.inet.tcp.sendbuf_auto=1
> > net.inet.tcp.sendbuf_max=16777216
> > net.inet.tcp.recvbuf_max=16777216