Tweet
Using the 64-bit command line client for the Linux BurnInTest (V3), when I attempt to run the test on multiple (about 25+) machines on the same network, sometimes some of the machines fail the test completely, giving an error that looks like the machine cannot talk at all to the remote host. The machines in question have both a 1 Gb Ethernet connection and a 10 Gb Ethernet connection, but only the 10 Gb connection exhibits this failure. Usually, after the failure rebooting and re-running the test on the machine(s) that failed allows it to pass. The length of time the test is run does not appear to be significant.
-
-
What sort of duty cycle were you using for the network test? Were you using a common destination for all the machines under test?
As the standard network test uses ICMP echo packets, which are not guaranteed to be delivered, it's always possible that the destination system is dropping the packets once it passes a certain threshold of traffic.
Comment