On Mon, Aug 30, 2004 at 07:16:37PM +0200, Daniel Eriksson wrote: +> S?ren Schmidt wrote: +> +> > Hmm, looks like you re triggering the throtteling code, that +> > will lead to catastophic failure as it tosses out interrupts +> > causing this: +> +> Yes, that is very likely to be the source of my problems. However, the root +> cause is still not clear to me. Why would I get an interrupt storm when +> geom_stripe tastes the discs when everything else works like a charm? When new provider is created it is given for taste to other classes. This interrupt storm could be a result of some kind of abuse from one of existing classes. Could you provide output of command: # sysctl -b kern.geom.conftxt | awk '{print $2}' | sort | uniq It will be good if you could try unload some classes (gvinum, etc.) before loading geom_stripe, just to be sure. Increasing kern.geom.stripe.debug could be also interesting. -- Pawel Jakub Dawidek http://www.FreeBSD.org pjd_at_FreeBSD.org http://garage.freebsd.pl FreeBSD committer Am I Evil? Yes, I Am!
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:09 UTC