[Beowulf] Diskess cluster dictribution

John Hearns hearnsj at googlemail.com
Mon Jan 13 03:39:54 PST 2014


Patrick,
you say "Pelican is not tolerant of any interruption is communication
with slaves."

Does this mean that you are seeing frequent interruptions in network
connections between cluster nodes?
Have a look at the system log files - are you seeing entreies which
say ethernet interfaces are goign down?
Or NFS servers are unavailable?
If so, you really need to take a look at the network hardware.

Can you describe how many cluster nodes you have, the network cards
you have and what type of network switch,
maybe the model number?



More information about the Beowulf mailing list