On 20-Apr-09, at 8:28 PM, [email protected] wrote: > what are the limits that you are running into (network bandwidth, > server load, packet loss, other??) >
Sometimes even when there is just a single udpcast session running, point to point, on the same physical switch, and very little else happening on that switch... well it just gets stupid. Drops packets until it aborts. The wrapper scripts retry a couple of times and usually get the image to go all the way on the second try. Sometimes manual intervention is needed and it takes 3-5 tries. Other times, same equipment, same switch, 12-15 machines can receive the exact same image using unicast (a seperate udpcast session on unique ports per sender/receiver ) and it works just fine. All dozen udp-send processes are running on the same image server. If the image is small, say of just the first 12G of the disk (first partition only) it works 99.99999% of the time first try, whether there is one or 15 different udpcast sessions running on that segment. Intermittent problems would cause me to pull my hair out but I don't have any left at this point. I have tweaked all manner of udpcast parameters, asked the network guys to look at the switch, etc. We are using HP2848 switches, HP DL360G4 and G5 machines. CentOS 5.2 for the image server side. _______________________________________________ Discuss mailing list [email protected] http://lopsa.org/cgi-bin/mailman/listinfo/discuss This list provided by the League of Professional System Administrators http://lopsa.org/
