-----BEGIN PGP SIGNED MESSAGE-----
On 11/21/2014 5:53 AM, Bjoern A. Zeeb wrote:
> On 21 Nov 2014, at 06:10 , Shawn Webb <latt...@gmail.com> wrote:
>> Looks like I’m getting a kernel panic on heavy work loads (poudriere
run with 10 build slaves on an Intel Core i7 box). Below is a link to an
imgur album of screenshots I took with my phone.
>> I can reproduce this quite easily simply by spinning up a new
poudriere run. Should I file a bug report or is reporting here all
>> Link to screenshots: http://imgur.com/a/h64uB
> Which exact revision/version of FreeBSD is this (SVN revision in case
I'm at r274745. The code in question was written in 2003. I've narrowed
down the behavior to OpenVPN with IPv6 causing this kernel panic.
My OpenVPN server dishes out IPv6 as well as IPv4 in routing mode and is
configured to use UDP (which is the default). OpenVPN's default
keepalive is 120 seconds. If the tunnel goes down within that 120
seconds and OpenVPN doesn't notice, the tun device is still active but
packages are simply queued rather then delivered. Eventually OpenVPN
re-establishes the tunnel while keeping the tun device open, reusing it
rather than destroying/recreating it. The kernel panic is happening
either while the tunnel is down (yet tun device still is open) or when
the tunnel is being brought back up. If I set the keepalive to a really
small value (I currently have it at two seconds), then the kernel panic
Really interesting behavior.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (MingW32)
-----END PGP SIGNATURE-----
email@example.com mailing list
To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"