Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 16 Feb 2015 11:51:37 +0100
From:      Fabian Keil <freebsd-listen@fabiankeil.de>
To:        freebsd-questions@freebsd.org
Subject:   Re: FreeBSD 10.1 crashing under load
Message-ID:  <4e9beab4.5e4e3963@fabiankeil.de>
In-Reply-To: <alpine.DEB.2.19.4.1502120004310.7303@trent.utfs.org>
References:  <alpine.DEB.2.19.4.1502120004310.7303@trent.utfs.org>

next in thread | previous in thread | raw e-mail | index | archive | help
--Sig_/.9/WSbak48Tb1gv4A5ASu=Y
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable

Christian Kujau <lists@nerdbynature.de> wrote:

> I have a Lenovo Ideapad S10 netbook running FreeBSD 10.1-p5 and=20
> ever since I switched from Linux Fedora to FreeBSD the system crashes=20
> regularly when the system is under a bit of load :-\
>=20
> This netbook is operated in a headless mode, so I'm not using Wifi,=20
> Bluetooth or other fancy hardware features this netbook might have, only=
=20
> the system's disk and ethernet.
>=20
> I've put a few more details below, but what I wanted to ask is: are there=
=20
> any flags to set in FreeBSD so that it's more chatty when it crashes?=20

Using a kernel with WITNESS and INVARIANTS sometimes results in
the problem being caught earlier.

> Right now the system just reboots but leaves nothing in the system log as=
=20
> to _why_ it crashed. I've configured a crash dump device as per [0] but n=
o=20
> crash dump is ever taken. I'm running with encrypted swap (via GELI),=20
> maybe crashdumping doesn't work with encrypted swap?

IIRC it's not officially supported but you can dump on the device
below the geli layer:

fk@r500 ~ $swapinfo=20
Device          1K-blocks     Used    Avail Capacity
/dev/ada0s1b.eli   2097152        0  2097152     0%
fk@r500 ~ $dumpon -l
ada0s1b

Saving the dump automatically after boot doesn't work (for me),
but "service savecore start" works after "swapoff -a".

Obviously such a configuration is not without risks, especially
if you forget to overwrite the dump dev after saving the core.

> When running, this spawns some ~40 java processes that can be quite
> active on both CPU and network & disk I/O. It uses a lot of memory too
> and the machine does only have 1GB of physical memory (+1GB swap) but even
> if the application would use too much memory I'd expect the application to
> crash, not the whole system. The following is from "top -b" right before
> the machine crashed:

Maybe you need to play with vm.overcommit for details see tuning(7).

Fabian

--Sig_/.9/WSbak48Tb1gv4A5ASu=Y
Content-Type: application/pgp-signature
Content-Description: OpenPGP digital signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iEYEARECAAYFAlThy7YACgkQBYqIVf93VJ3qjACdFUjpm3Yx/j5TNxHYukEzUI/j
7jQAoLcBdQAc4TnHOmfBdTuSJH6ZQwtP
=+7wm
-----END PGP SIGNATURE-----

--Sig_/.9/WSbak48Tb1gv4A5ASu=Y--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4e9beab4.5e4e3963>