[Info-vax] PEA0: Errors

[Volker Halle]

Christoph,

if the problem manifests itself as connect lost/re-established
connection messages and you have no signs of link lost (MC LANCP SHOW
DEV/INT shows the most recent LAN driver console messages) or other
LAN-related transmit/receive errors (MC LANCP SHO DEV/COUNT), you
should concentrate on possible multicast traffic problems !

To maintain cluster connectivity, every node sends a Cluster Hello
multicast message every 3 seconds. If it does not receive such a
message from another known cluster member within about 9 seconds, it
declares a 'connection lost' event. If it then receives a cluster
hello message from that node, it logs a 'connection re-established'
event. If no messages is recieved from the other node within
RECNXINTERVAL seconds, that connection is timed out and the node is
removed from the cluster. If then another clsuter hello message from
that node gets through, it has to take a CLUEXIT crash to get into the
cluster again.

If you carefully look at these messages in OPERATOR.LOG and find out,
which node looses connection to which other node, you may be able to
isolate this problem to a certain port or switch.

Volker.