I can reliably produce a master node panic. The
resulting console message is:
Process 4126(te), uid=6005, killed because it received
an unrecognized message type (0x0) from the remote site
where it was running
DEPUTY: Unexpected non-urgent request type 0
Kernel panic: deputy_wait
This occurs when I compile my program("te") with profiling
enabled. The compiler is gcc v-3.2.3. The master OS is
RHEL 3 AS
with kernel 2.4.26 and the 2.4.26-om1 patch applied.
The program runs fine for 1-2 minutes then the master
node panics only producing the message I listed above.
If it matters, the nodes are modified versions of the
chaos distribution that run the official 2.4.28-om1 kernel.