On Wed, Dec 16, 2009 at 07:42:34 -0600, Jay Faulkner wrote: > On Wed, Dec 16, 2009 at 22:43:44 +1100, Simon Horman wrote: > > On Wed, Dec 16, 2009 at 10:07:33 +0200, Aleksey Chudov wrote: > > > I have 2.6.31.2 kernel and it seems like code was changed and there is no > > > msleep_interruptible in net/netfilter/ipvs/ip_vs_sync.c > > > > Completely untested, but I think that this will have the > > effect that you are after. > > > > diff --git a/net/netfilter/ipvs/ip_vs_sync.c > > b/net/netfilter/ipvs/ip_vs_sync.c > > index e177f0d..20b05db 100644 > > --- a/net/netfilter/ipvs/ip_vs_sync.c > > +++ b/net/netfilter/ipvs/ip_vs_sync.c > > @@ -757,7 +757,7 @@ static int sync_thread_master(void *data) > > ip_vs_sync_buff_release(sb); > > } > > > > - schedule_timeout_interruptible(HZ); > > + schedule_timeout_interruptible(HZ/10); > > } > > > > /* clean up the sync_buff queue */ > > How does this work with people using the NoHZ kernels?
Hi Jay, I believe that the code will work as expected on NO_HZ=y kernels. This is because the timer will be scheduled as needed. My understanding being that the difference that NO_HZ=y gives is that the kernel won't wake up at a clock-tick if no timers are set in that slot. _______________________________________________ Please read the documentation before posting - it's available at: http://www.linuxvirtualserver.org/ LinuxVirtualServer.org mailing list - [email protected] Send requests to [email protected] or go to http://lists.graemef.net/mailman/listinfo/lvs-users
