Tomas,

I have at least your problems #1 and #2, and found fixes for both:

After installing ganglia and php 7.2 (for me, on Ubuntu 20) as described in
       
https://websiteforstudents.com/how-to-install-ganglia-monitoring-server-on-ubuntu-18-04-16-04/
<https://websiteforstudents.com/how-to-install-ganglia-monitoring-server-on-ubuntu-18-04-16-04/>

I found that the main cluster view was *blank*, with just the UI buttons
but no individual hosts nor any cluster-aggregate graphs.
But the "physical view" button did work, and so did the menu choosing
individual hosts in the cluster -- the per-host display looked normal,
fully populated with the expected graphs.

The web server error log had a clue:

Mon Dec 28 12:27:38.491219 2020] [php7:error] ... PHP Fatal error: 
Uncaught Error: [] operator not supported for strings in
/usr/share/ganglia-webfrontend/cluster_view.php:34\nStack trace:\n#0
/usr/share/ganglia-webfrontend/cluster_view.php(757):
get_picker_metrics(Array, Array, '/usr/share/gang...', 'rrdtool')\n#1
/usr/share/ganglia-webfrontend/index.php(51):
include_once('/usr/share/gang...')\n#2 {main}\n  thrown in
/usr/share/ganglia-webfrontend/cluster_view.php on line 34

It turned out that cluster_view.php initialized a variable as a string,
then assigned to it as an array, like
   $context_metrics = "";
   [...]
   $context_metrics[] = $report_name;

Thanks to this stackoverflow thread
<https://stackoverflow.com/questions/5879675/fatal-error-operator-not-supported-for-strings>
for explaining that this used to work in php5, but doesn't in php 7.1
and later.
This patch fixed it:


--- cluster_view.php.ORIG    2014-04-04 05:20:44.000000000 -0500
+++ cluster_view.php    2020-12-28 18:52:47.936940947 -0600
@@ -23,7 +23,7 @@
 }
 
 function get_picker_metrics($metrics, $reports, $gweb_root,
$graph_engine) {
-  $context_metrics = "";
+  $context_metrics = [];    // for php 7.1+ we must initialize this to
an array for $var[] = ... assignment to work.  Thanks to:
https://stackoverflow.com/questions/5879675/fatal-error-operator-not-supported-for-strings
   if (count($metrics)) {
     foreach ($metrics as $host_metrics) {
       foreach ($host_metrics as $metric_name => $metric_value) {


Also, the Ubuntu ganglia package didn't contain
/var/lib/ganglia/conf/*.json configuration entries.  I copied those from
another ganglia installation (part of the Rocks 6 cluster
distribution).   They might be helpful as examples.

/var/lib/ganglia/conf/default.json  contains:

{
    "included_reports":
["load_report","mem_report","cpu_report","network_report"]
}

/var/lib/ganglia/conf/view_default.json  contains:

{"view_name":"default","items":[],"view_type":"standard"}



On 5/23/20 7:08 AM, Tomas Aronsson wrote:
> Hi!
>
> My name is Tomas! Just installed ganglia a week ago to monitor my
> little raspberry pi cluster. I have three issues:
> 1) I cannot see any cluster summary graphs, only the host-specific
> ones are populated. Is this normal? How can I see the summary graphs?
> Summary empty: https://emyro.se/ganglia_fail1.PNG
> <https://emyro.se/ganglia_fail1.PNG>
> Specific host not empty: https://emyro.se/ganglia_fail2.PNG
> <https://emyro.se/ganglia_fail2.PNG>
>
> 2) The hostnames are just IP addresses. Is there anyway to get the
> hostname? For the hosts I've set the hostname, but only the IP show up?
> https://emyro.se/ganglia_fail3.PNG <https://emyro.se/ganglia_fail3.PNG>
>
> 3) I get some very strange readings from the raspberry pi hosts. It
> says network traffic is spiking at PB/s (lol!) eventhough it's more
> like MB/s. Any ideas?
> https://emyro.se/ganglia_fail4.PNG <https://emyro.se/ganglia_fail4.PNG>
>
> Any help is greatly appreciated!
>
> I use the following settings:
> main node 192.168.22.100 (neptune)
> raspberry pi 1 ( 192.168.22.102)
> raspberry pi 2 ( 192.168.22.103)
>
> My syslog contains no error messages.
>
> gmond.conf (same for all hosts):
>
> cluster {
>   name = "my cluster"
>   owner = "unspecified"
>   latlong = "unspecified"
>   url = "unspecified"
> }
> udp_send_channel {
>   host=192.168.22.100
>   port = 8649
>   ttl = 1
> }
> udp_recv_channel {
>   port = 8649
> }
>
> gmetad.conf:
> data_source "my cluster" 15 192.168.22.100:8649
> <http://192.168.22.100:8649>
>
>
> _______________________________________________
> Ganglia-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/ganglia-general

_______________________________________________
Ganglia-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ganglia-general

Reply via email to