You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Running Ganglia 3.7.2 on CentOS 8, we discovered that the node graphs in the cluster view section of a ganglia cluster report would go blank after an auto-refresh. On load, the graphs would normally show a CPU or network report like this:
then after an auto-refresh, all graphs would be empty, although high load nodes would still have a colored frame, like this:
Initial investigations suggested that graphs fetched after the update were missing a value for the "g" parameter, which specifies the metric to graph. A lot more troubleshooting found that the cluster_refresh.tpl template was not getting the metric_name variable properly set.
I modified cluster_view.php and moved what is currently line 636, $tpl_data->assign("metric_name", "{$user['metricname']}");, up to above line 608, if (! $refresh) {, so the variable would be set for both refresh and non-refresh cases. This seems to fix the issue.
I am not offering a means to recreate this issue, so I encourage others to speak up if they have a similar problem.
With that, I leave this issue to the Ganglia Gurus to handle as they see fit.
The text was updated successfully, but these errors were encountered:
I've the same issue, running on CentOS 7.9, after upgrade to 3.7.5, all the graph shows "Empty RRDtool command, Likely bad graph config", previously it was fine (older version)
Running Ganglia 3.7.2 on CentOS 8, we discovered that the node graphs in the cluster view section of a ganglia cluster report would go blank after an auto-refresh. On load, the graphs would normally show a CPU or network report like this:
then after an auto-refresh, all graphs would be empty, although high load nodes would still have a colored frame, like this:
Initial investigations suggested that graphs fetched after the update were missing a value for the "g" parameter, which specifies the metric to graph. A lot more troubleshooting found that the cluster_refresh.tpl template was not getting the metric_name variable properly set.
I modified cluster_view.php and moved what is currently line 636,
$tpl_data->assign("metric_name", "{$user['metricname']}");
, up to above line 608,if (! $refresh) {
, so the variable would be set for both refresh and non-refresh cases. This seems to fix the issue.I am not offering a means to recreate this issue, so I encourage others to speak up if they have a similar problem.
With that, I leave this issue to the Ganglia Gurus to handle as they see fit.
The text was updated successfully, but these errors were encountered: