I still haven't seen an explanation as to why some of the Roulette testbed nodes on the 128.32.130.0/25 were making requests to NTP servers outside Berkeley. But we have recieved some further email from the contacts for the Roulette sensors and the contacts for the 128.32.130.0/25 network.
********begin some forwarded excerpts****************** roulette contact: "Not quite sure why only a subset of the nodes report the problem, the image on all of them are the same. Also, I used to manually set the dates and times, never used NTP. Is there some way to check who has logged onto the nodes?" 128.32.130.0/25 network contact: "I have blocked outbound traffic from all the roulette-*.cs.berkeley.edu nodes on the 128.32.130.0/25 network." 128.32.130.0/25 network contact: "The Soekris nodes of the roulette testbed are just small Linux machines. It's likely that the roulette Linux install may be misconfigured to use off-campus NTP servers. The (128.32.130.0/25 network border) firewall was blocking inbound traffic to 128.32.130.0/25 except from a few select hosts. Unfortunately because the firewall was originally configured with only motes in mind, there was no such restriction on outbound traffic. The misconfigured roulette nodes were probably requesting NTP updates and not receiving responses. The firewall rules have also been updated." roulette contact: "It'll take some time before I can grab hold of a node and find out what's wrong. In the meanwhile none of the nodes will be up." ********end some forwarded excerpts***************** mark kraitchman (for [EMAIL PROTECTED]) _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
