> i think the current code also allows for gmetad hierarchies to do the summary > right, because 3.1 will use (unless I misunderstood the code) float types for > memory metrics anyway, but haven't yet test that.
> do you have any probe otherwise? Imagine that I have many grids around the globe and then I have a worldwide overview gmetad that polls all those grids. It is not practical to upgrade all grids simultaneously. So, what happens when some grids change to double but others remain on integers? Since you have applied my type-merge patch to gmetad, we *might* be OK in the RRDs; I'm still looking for a test case to show otherwise. But, look at what happens with the /?filter=summary mode: Current trunk: # echo "/?filter=summary" | netcat localhost 47001 | grep mem_total <METRICS NAME="mem_total" SUM="1660944384" NUM="162" TYPE="uint32" UNITS="KB" SLOPE="zero" SOURCE="gmond"/> <METRICS NAME="mem_total" SUM="489729520" NUM="60" TYPE="uint32" UNITS="KB" SLOPE="zero" SOURCE="gmond"/> <METRICS NAME="mem_total" SUM="882306972" NUM="102" TYPE="double" UNITS="KB" SLOPE="zero" SOURCE="gmond"/> Trunk with complete bug#128 patch applied: # echo "/?filter=summary" | netcat localhost 47000 | grep mem_total <METRICS NAME="mem_total" SUM="1372036492" NUM="162" TYPE="double" UNITS="KB" SLOPE="zero" SOURCE="gmond"/> <METRICS NAME="mem_total" SUM="489729520" NUM="60" TYPE="double" UNITS="KB" SLOPE="zero" SOURCE="gmond"/> <METRICS NAME="mem_total" SUM="882306972" NUM="102" TYPE="double" UNITS="KB" SLOPE="zero" SOURCE="gmond"/> Correct answer: # dc -e '489729520 882306972 + p' 1372036492 Maybe we are OK if nothing ever uses the summary filter. But I prefer it to be correct. -twitham ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Ganglia-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-developers
