Thanks for the tips. The attached tar.gz file is the result of adding a call to phpinfo() at the top of graph.php and switching on the debug flag. At the bottom is the command that is being run to try to generate a graph for one of the nodes that is actually in a different cluster. Here is the command.
/usr/local/rrdtool/bin/rrdtool graph - --start 1194960576 --end 1194964176 --width 800 --height 600 --lower-limit 0 --rigid --title 'quad001.beowulf.cluster Load last hour' --vertical-label 'Load/Procs' DEF:'load_one'='/var/lib/ganglia/rrds/NEMO cluster @ POL/quad001.beowulf.cluster/load_one.rrd':'sum':AVERAGE DEF:'proc_run'='/var/lib/ganglia/rrds/NEMO cluster @ POL/quad001.beowulf.cluster/proc_run.rrd':'sum':AVERAGE DEF:'cpu_num'='/var/lib/ganglia/rrds/NEMO cluster @ POL/quad001.beowulf.cluster/cpu_num.rrd':'sum':AVERAGE AREA:'load_one'#CCCCCC:'1-min Load' LINE2:'cpu_num'#FF0000:'CPUs' LINE2:'proc_run'#0000FF:'Running Processes' There is no directory named "quad001.beowulf.cluster" in "/var/lib/ganglia/rrds/NEMO cluster @ POL". This explains why the command fails leading to a blank space on the page. For the nodes that are correctly listed as being at POL it is difficult to find out why the wrong information is being displayed. Can I trust Ganglia to display the correct information from the rrds files or is there another way to find out what they contain? One clue is in the information being shown for node025 at POL. This node is being shown as down on the POL cluster report, but the node is in fact alive and well. The output from gmond at POL shows the correct information and the rrds files for this node are being updated every minute as you would expect. This suggests to me that the rrds files for the POL cluster do contain the right information. The rrds files for the BAS cluster are in directory "/var/lib/ganglia/rrds/ClusterVision Cluster/node025.beowulf.cluster". They have not been updated for the past few days, as you would expect for a node that is down. So it seems that Ganglia is getting at least some of the information from the wrong rrds files even though the paths in the rrdtool commands are correct. I suppose the next step is to debug the other php scripts involved. Where is the best place to start? -Dan. On Monday 12 Nov 2007 18:59, aurbain wrote: > flick on the debug flag in graph.php, and redraw the page. You should > see the command line which would get run to make the graph. Copy and > run > test.png and take a look at it. > -- Mr. D.A. Bretherton Environmental Systems Science Centre Harry Pitt Building 3 Earley Gate Reading University Reading, RG6 6AL UK Tel. +44 118 378 7722
quad001.graph.php.html.tar.gz
Description: application/tgz
------------------------------------------------------------------------- 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-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-general
