On Sun, 2007-06-17 at 15:23 -0400, Patterson, Ron (NIH/NLM/NCBI) [C] wrote:
> For example, if I use the following: > > /usr/bin/gmetric --conf=/etc/gmond.conf -nmystat -v1.0 -tfloat -uunits > > Run every minute from all 200 hosts in my cluster, I will see gmond > start growing in size almost immediately. Have you tried looking at the metrics stored in memory on a single host? You can just run something like "telnet localhost 8649" to get a dump of all the XML data. The reason I suggest this is to get an idea as to what is causing the memory consumption. If you find that the size of the metric data stays roughly the same, then that would seem to point to a memory leak issue. If you found that the size of the metric data kept growing (and hence was responsible for using up the memory), then maybe something weird is happening and gmond is somehow treating each new incoming data point as a new metric. Then it would keep accumulating values in memory instead of overwriting the previous value. Of course, your example above seems pretty straight forward so I would be surprised if the latter were true. I'm betting that your diagnosis of a memory leak is more likely correct. This simple test would at least rule out the second possibility. -- Rick Mohr Systems Developer Ohio Supercomputer Center ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Ganglia-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-general
