[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: more bgpd weirdness
- To: misc_(_at_)_openbsd_(_dot_)_org
- Subject: Re: more bgpd weirdness
- From: Arvid Grøtting <arvidg_(_at_)_netfonds_(_dot_)_no>
- Date: Sat, 14 Aug 2004 11:59:23 +0200
- Cancel-lock: sha1:IwK6zNKmePpemA4ggMxU+bCsLTM=
- Mail-copies-to: never
- Mail-followup-to: misc_(_at_)_openbsd_(_dot_)_org
- Organization: No such thing.
Arvid Grøtting <arvidg_(_at_)_netfonds_(_dot_)_no> writes:
> I'm suspecting ste(4) and the D-Link 580 card here, and I'll try
> swapping it for a spare 570 shortly. Stay tuned.
Swapping the NIC does indeed give a much better behaviour under
network load, or at least so it seems.
This means that the ste(4) driver must have regressed; the exact same
box worked quite nicely under heavy network load with OpenBSD 3.2 or
3.3 (I don't quite remember which at the moment) and with my
simplistic multicast reception patch (the one that doesn't care about
the filter, but simply turns on the all-multicast-packets-are-belong-
to-me bit, posted here or in -tech a while ago).
Now ste(4) multicast filtering works with the 580, but it freaks out
(approaching 100% interrupt load) at any sight of actual network use.
I haven't been able to try with an up-to-date bgpd yet, though. The
(same snapshot) bgpd still gives up if I try to let it reload the
config file, so that's a separate problem. I haven't retried the
"default config" config file, either, yet.
Visit your host, monkey.org