On Wed, Oct 31, 2007 at 04:29:21PM -0600, Brad Nicholes wrote:
> -* Remove the 4T limit on ganglia metric results
> -* Modify all byte count metric to 8 bytes ints

moving to 8 byte ints won't solve the problem, will only push it further and
will make the memory utilization for gmond and disk utilization for gmetad
double for these specific metrics.

the increase on memory and disk cost could have a bigger impact with gmond
scalability for very large clusters.

from my experience when you go over 500 nodes multicast is prohibitive and 
unicast will require sub segmenting the nodes in between multiple collectors
to avoid having gmond eating"significant" amounts of memory.  With over
1000 nodes gmetad needs to run in a ramdrive and for that the size of the RRD
better be smaller than your RAM.

As I mentioned before, one of the "strengths" from gmond IMHO was that it used
the "right" multiplier on each case to try to keep the memory/bandwith/storage
costs as low as possible, making that multiplier configurable so that the
definition of "right" is done by the user would be the only sensible approach
AFAIK.

That should allow them to define memory/disk/bandwith in bits, bytes, KB, MB,
GB or whatever granularity fits them better.

Carlo

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Ganglia-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ganglia-developers

Reply via email to