On Thu, Jul 2, 2009 at 05:29, <[email protected]> wrote: >> 2. Watch the network traffic between B and A, and see if B is actually >> sending new_metric to A or to a multicast that A can see >> and looks at. > > Any advice on how I do that ? I've done a tcpdump on host a but I can't > reconstruct a sensible XML tree from that output.
Use the "-d <level>" option for both gmond and gmetad. Valid numbers are "1" and higher. I think that level 2 or 3 (maybe 4) will start dumping messages about specific metrics generated (gmond) and received (gmond and gmetad). > Is there any setting I can feed to gmond on node A box that says 'Log all > incoming data somewhere' ? > >> 3. Restart both gmond's, and check the syslogs on both hosts for any >> messages from gmond. > > I've done that and there is nothing in the Unix /var/log/messages file on > node A and nothing in event viewer for node B. The only messages I am seeing > are startup and shutdown. I've never seen, that I recall, any part of Ganglia make use of syslog. However, the debug mode I just mention dumps to STDERR, so you can capture that, and dump to a log. -- Jesse Becker GPG Fingerprint -- BD00 7AA4 4483 AFCC 82D0 2720 0083 0931 9A2B 06A2 ------------------------------------------------------------------------------ _______________________________________________ Ganglia-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-general
