considering your msg about rst etc -- there are a number of potential failure modes including the most likely one below: + because of the extra delay, the initial syn handshake takes too long to complete causing way too many entries in the listen queue, so this saturates the server side. cheers luigi On Tue, Apr 27, 2004 at 02:25:34AM +0200, Ivan Voras wrote: > Luigi Rizzo wrote: > > are you sure you aren't running out of mbufs ? > > netstat -m should tell you. > > No, its not mbufs (such messages are shown on the console I believe and > I haven't seen them; also, running netstat -m states there's plenty of > buffers remaining). > > > Additionally, note that the 250ms of delay are probably way too much > > for your config (you are adding 250ms each way, which makes a 500ms RTT, > > not accounting for transmission times -- i think normal values here > > are more like 150-200ms rtt), and possibly irrelevant given how > > large your queues are -- a full-size packet is 12000 bits or 200ms, > > you can have up to 20 queued... > > Oh sorry, I pasted from a wrong shell script. I've been experimenting > with different values and am considering settling at 75ms, queue of 5. > I'm still not sure: how does the number of buckets influence the operation? > > I don't think it's clear to me how would such large queues produce my > errors (connection reset by peer & broken pipe). Is this reasoning > correct: in the above case, 200ms*20 = 4s, so in the worst case the > packet from the end of a queue will travel 4s until it reaches its > destination. Is 4s enough for a timeout of some sort? > > > I've tried running on a different machine, running 4.9-release, and > there I also get this error (in large numbers): > Error: socket: address is unavailable.: Can't assign requested address > > > -- > C isn't that hard: void (*(*f[])())() defines f as an array of > unspecified size, of pointers to functions that return pointers to > functions that return void. > _______________________________________________ > freebsd-current_at_freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-current > To unsubscribe, send any mail to "freebsd-current-unsubscribe_at_freebsd.org"Received on Tue Apr 27 2004 - 02:32:57 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:52 UTC