On 2013-05-15, Peter Bisroev <[email protected]> wrote: > On Tue, May 14, 2013 at 4:55 AM, Stuart Henderson <[email protected]> > wrote: >> Anyone have any ideas about how to improve TCP performance with huge >> numbers of out-of-order packets? >> >> 62653661 packets received >> 25373283 acks (for 43239433893 bytes) >> 2225419 duplicate acks >> 20139430 packets (21139432159 bytes) received in-sequence >> 989606 completely duplicate packets (299125194 bytes) >> 51753 old duplicate packets >> 362 packets with some duplicate data (144255 bytes >> duplicated) >> 15927761 out-of-order packets (19170915512 bytes) >> 28812 packets (28812 bytes) of data after window >> 28812 window probes >> 259673 window update packets >> 38231 packets received after close >> 21 discarded for bad checksums >> 26790492 packets hardware-checksummed >> > > Hi Stuart, > > Out of curiosity, what kind of link are you getting this on?
Two pairs of per-packet load balanced (slow) ADSLs, round robinning connections between the pairs to avoid *too* many problems. > What kind of hardware? R210 II E31220L (pretty much idle). > I ask because I had something similar with a few of my Debian boxes. > The problem was related to non-pegged interrupts on the E5-2600 Xeons. > > Cheers, > --peter
