There is something odd going on. Your server has been "dead" for 42
seconds during this run... Several lost packets (TCP?) on your ethernet?
[grabs stevens] Retransmits should be occurring at 1,3,6,12 and 24 second
intervals after the first packet hits the bucket. That adds up to 46
seconds.
I don't understand.
Well, the clients are old BSD4.3 network stack SunOS boxes, the
10baseT network in question is pummeled over a buggy switch with
gigabytes of researcher data set NFS activity from 20 compute cluster
client SS20's running 35 or so batch compute jobs, plus 15 or so
personal workstations.
I'm sure the 42 second response can be accounted for, it only occurs
over the 10baseT network, it never happens when I do a run over a
direct 100baseT pipe.
---------------------------------------------////
Yow! 11.26 MB/s remote host TCP bandwidth & ////
199 usec remote TCP latency over 100Mb/s ////
ethernet. Beat that! ////
-----------------------------------------////__________ o
David S. Miller, davem@caip.rutgers.edu /_____________/ / // /_/ ><