On Wed, Jul 16, 2008 at 04:25:13PM -0400, Jason A. Smith wrote: Hi Jason, glad to see you are still around hacking on ganglia
> A while ago I submitted some patches to bugzilla that we use in our > ganglia, but some have not been added to the main distribution. I'd adopted some of your 3 year old bug reports and will try to get them resolved > Do > others find these useful? If so could they be added? I also have a few > new/updated patches. Sorry, but they are against the old 3.0.7 stable > version, not the new 3.1.x, since that is the only place I can test them > right now. I am pretty sure they are useful, as they are working for you at least and other users might have the same problems that you were trying to solve with them. All we need to be sure is that the solution is generic enough to apply for all cases, and that it can be applied and solves the problem in trunk, as explained in : http://ganglia.wiki.sourceforge.net/ganglia_works Indeed, unless I am confusing you with someone else, you have the commit rights to do it yourself. some comments about your patches below : > 2. http://bugzilla.ganglia.info/cgi-bin/bugzilla/show_bug.cgi?id=37 > > With ganglia 3.0.7 I am still seeing "0 - 0" as the vertical label on > the cluster view for all of the node plots. As I explained in the bug > report, for the version of rrdtool that I am using (now 1.2.27) the > PRINT command doesn't seem to work when the output filename is set to > '-', but it does work with an empty filename ''. I also updated this > patch for ganglia version 3.0.7. Patch does apply to trunk cleanly, but > is untested. I am not sure about this, as at least in my setup while testing in the command line the commands generated the min/max are getting set correctly (using rrdtool 1.0.40) and removing the filename only generates an error. When tested with rrdtool 1.2.27 the min/max values are not being generated as you reported and as described in the man page (all output is suppressed), using /dev/null instead as the filename to be generated seems to have the desired effect, but is an expensive operation if all that was needed was to get the min and max values for the vertical label. It seems also that Jesse hijacked your report with an alternative or probably related fix for the vertical label, which is since also obsolete because trunk and 3.1 moved to the modular graph setup and so this report needs probably some more work/thinking to integrate cleanly with trunk. > 3. http://bugzilla.ganglia.info/cgi-bin/bugzilla/show_bug.cgi?id=36 > > I guess this patch never made it into the 3.0.x series. It has been proposed for backport for almost 2 months now, but seems no other developer has been so far interested on voting it, and therefore as per the RTC rules that apply to 3.0 it is not yet backported there. > I assume from > the updates made a few months ago that this patch is in the new 3.1.x > and the 3.0.x series is dead now? Fix is in 3.1, 3.0 is not dead and will have at least a 3.0.8 released sometime soon which could include this fix. > Note, with the revised patch, when a > node is down and you look at the node view, the message displayed by the > template is: "This cluster is down." if it is doing that, it is a bug, as that message should be only shown if the cluster where that node is indeed down (for example you are looking at an unicast collector that is no longer running). if gmetad is still receiving updates from that cluster, then it will be able to use the last cluster update to know relatively to that cluster for how much time the node has been down. the message is there to explain why the length of time a node has been down doesn't increase if also all other nodes (from the perspective of gmetad) from that cluster are also down, as there is no time to use for reference. > Shouldn't that be "This node is > down.", since you are only looking at a single node? that shouldn't be needed, as it is implied by the "down" status for the node everywhere else (including the color and icon used). Carlo ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Ganglia-developers mailing list Ganglia-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ganglia-developers
