Telegraf > Procstat > E! Error in plugin [inputs.procstat]

Telegraf
,
1

==Configuration==

OS:

  • CentOS Linux release 7.4.1708 (Core)

PKGs:

  • influxdb-1.3.7-1.x86_64
  • grafana-4.6.2-1.x86_64
  • telegraf-1.4.3-1.x86_64

==Telegraf Configuration== /etc/telegraf/telegraf.conf
[[inputs.procstat]]
exe = "memcached"
prefix = “memcached”
[[inputs.procstat]]
exe = "ntpd"
prefix = “ntpd”

==Telegraf Output==

  • Plugin: inputs.procstat, Collection 1
    2017-12-12T18:49:22Z E! Error in plugin [inputs.procstat]: E! Error: procstat getting process, exe: [ntpd] pidfile: [] pattern: [] user: [] Error running /bin/pgrep: exit status 1
  • Plugin: inputs.procstat, Collection 1
    2017-12-12T18:49:19Z E! Error in plugin [inputs.procstat]: E! Error: procstat getting process, exe: [memcached] pidfile: [] pattern: [] user: [] Error running /bin/pgrep: exit status 1

***I’ve tried also in /etc/telegraf/telegraf.conf to have the below config but no graphs in Grafana then.

[[inputs.procstat]]
exe = "telegraf|.*"
prefix = “pgrep_svcs_process”

VS

[[inputs.procstat]]
exe = "telegraf|.*"
prefix = “pgrep_serviceprocess”

!!!
The point is to modify the /etc/telegraf/telegraf.conf and to collect all information for all processes and after this to be able to have the metrics in Grafana.

2

Is memcached running, what do you get when you run pgrep memcached?

3

*** pgrep ntpd**
*** pgrep memcached**

doesn’t return nothing and it’s clear that is a normal behavior of telegraf if a process is not up.

What are your recommendation in order to not have multiple definitions in “/etc/telegraf/telegraf.conf” like below:
[[inputs.procstat]]
exe = "memcached"
prefix = “memcached”
[[inputs.procstat]]
exe = "ntpd"
prefix = “ntpd”
.
.
.
etc

—===—===—
could I use something else like below for all processes:
[[inputs.procstat]]
exe = "telegraf|.*"
prefix = ?

What to do for the processes where “pgrep process” will not reveal nothing like: pgrep kibana, pgrep elasticsearch?
If I’ll try “pgrep -u root, kibana; pgrep -u root, elasticsearch” then will be output there for process.
All apps are installed under root permission…

4

I would use the former, though generally I prefer:

[[inputs.procstat]]
  exe = "memcached"
  name_override = "memcached"

I’ve never used telegraf|.*, I looks like it would just match everything in which case it seems simpler to just use exe = ".*". If you collect all the processes in one plugin then it won’t be helpful to set a prefix. There is another more important issue with collecting all processes though: in order to avoid conflicts you must use pid_tag = true and due to the number of pids created on a normal system you will end up creating many series which add lot’s of strain on the Database. You might be able to solve this with a short retention policy, but you are better off monitoring a limited number of long running processes with this plugin.

1 Like
5

Thank you for your approach.
One little mention because I’m searching a solution and maybe you have a better suggestion.

For example if I want to collect metrics with Telegraf for:

  • node_exporter
  • Kibana

/etc/telegraf/telegraf.conf
[[inputs.procstat]]
exe = "node_exporter"
name_override = “node_exporter”
[[inputs.procstat]]
exe = "kibana"
name_override = “kibana”

I have a question that for the moment I don’t have a solution.
By default Telegraf will do a $ pgrep kibana, $ pgrep node_exporter but the thing is that if you have the below situation then it will be not relevant.

[root@server ~]# systemctl status kibana
● kibana.service - Kibana
Loaded: loaded (/etc/systemd/system/kibana.service; enabled; vendor preset: disabled)
Active: active (running) since Tue 2017-12-12 21:42:42 EET; 11h ago
Main PID: 4206 (node)
CGroup: /system.slice/kibana.service
└─4206 /usr/share/kibana/bin/…/node/bin/node --no-warnings /usr/share/kibana/bin/…/src/cli -c /etc/kibana/kibana.yml

[root@server ~]# systemctl status node_exporter
● node_exporter.service - Node Exporter
Loaded: loaded (/etc/systemd/system/node_exporter.service; enabled; vendor preset: disabled)
Active: active (running) since Thu 2017-12-07 17:24:17 EET; 5 days ago
Main PID: 581 (node_exporter)
CGroup: /system.slice/node_exporter.service
└─581 /root/Prometheus/node_exporter-0.15.0.linux-amd64/node_exporter

[root@server ~]# pgrep node
581
4206
[root@server ~]# pgrep node_exporter
581

***How to colect metrics for Kibana only?

Solution could be below?
===========PID file?============
[[inputs.procstat]]
** exe = “kibana”**
** name_override = “kibana”**
** pid_file = “/var/run/kibana.pid”**

6

That should work, also in 1.5.0 (being released very soon) you will be able to select all the processes for a systemd service file:

[[inputs.procstat]]
  systemd_unit = "kibana.service"
  name_override = "kibana"
1 Like
7

Many thanks, I’ll implement in this way.

8

E! Error parsing /etc/telegraf/telegraf.conf, line 1937: field corresponding to systemd_unit' is not defined in*procstat.Procstat’

It seems that the telegraf will not start.

9

Make sure you have Telegraf 1.5.0-rc2 or newer if you want to use this option, the final version will be released soon.

1 Like