web at 3dresearch.com
Wed Mar 29 05:06:57 UTC 2017
On Wed, 29 Mar 2017 04:45:28 +0200
Polytropon <freebsd at edvax.de> wrote:
> On Tue, 28 Mar 2017 16:06:34 -0400, Janos Dohanics wrote:
> > The system is FreeBSD 11.0-STABLE #0 r314885 amd64. It is mostly
> > used to run Cyrus, Postfix, Amavisd, Clamd. Kernel is not
> > customized, except for "ident".
> > Could this problem be related to the additional swap space provided
> > by a swapfile?
> > # cat /etc/fstab
> > # Device Mountpoint FStype Options
> > # Dump Pass#
> > /dev/ada0p2 none swap sw
> > 0 0 /dev/ada0p3 / ufs rw
> > 1 1 /dev/ada0p4 none swap
> > sw 0 0 md0 none
> > swap sw,file=/swapfile,late 0 0
> > # swapinfo -hm
> > Device 1M-blocks Used Avail Capacity
> > /dev/ada0p2 256 241M 15M 94%
> > /dev/ada0p4 750 290M 460M 39%
> > /dev/md0 4096 289M 3.7G 7%
> > Total 5102 820M 4.2G 16%
> > I'd appreciate your advice.
> Do you have statistics about your CPU and I/O load? Both can
> cause a system to become unresponsive. Using swap will start
> in case the RAM is "full", and because you have three mechanisms
> of swap on the same disk (ada0), this could be the reason: You
> have two swap partitions and a swap file, all of them residing
> on the same disk, and all of them are in use, so that could be
> the reason for I/O load...
I have no statistics beyond what I have in the logs (maillog,
amavisd.log, messages), plus I have tried to keep an eye on top(1). I
did notice that at times CPU utilization, RAM and swap use (before
adding the 4 GB md(4) device as the attempted remedy) were close to
100%, while mailq(1) kept growing.
You are quite right, I should have thought about I/O load, and I
didn't. I'll do some experimentation.
In the meantime, Google turned up this page which looks relevant to the
error message in my original post:
Could you please explain?
More information about the freebsd-questions