Telegraf: ERROR in input [exec]: Errors encountered: [exec: exit status 1 for command '/opt/temp.py']

[roycepope]

Bug report

I am trying to collect temperature and humidity data from my raspberry pi. The telegraf log keeps spitting out the following error:
ERROR in input [exec]: Errors encountered: [exec: exit status 1 for command '/opt/temp.py']

I am able to run the script standalone and get the exit status 0. I tried running telegraf in debug mode but it produces the same error.

The response data from the script returns pretty much immediately and is definitely under the defined timeout value.

Relevant telegraf.conf:

[agent]
interval = "10s"
round_interval = false
metric_batch_size = 1000
metric_buffer_limit = 10000
collection_jitter = "1s"
flush_interval = "10s"
flush_jitter = "0s"
precision = "ms"

[[inputs.exec]]
commands = ["/opt/temp.py"]
timeout = "9s"
data_format = "influx"

System info:

Telegraf (version 1.0.0)
Linux rpi 4.4.21-v7+ #911 SMP Thu Sep 15 14:22:38 BST 2016 armv7l GNU/Linux

Expected behavior:

Actual behavior:

/var/log/telegraf.log
2016/10/07 04:02:45 Starting Telegraf (version 1.0.0)
2016/10/07 04:02:45 Loaded outputs: influxdb
2016/10/07 04:02:45 Loaded inputs: cpu disk diskio kernel mem processes system ping exec net nstat
2016/10/07 04:02:45 Tags enabled: host=rpi
2016/10/07 04:02:45 Agent Config: Interval:20s, Debug:false, Quiet:false, Hostname:"rpi", Flush Interval:10s
2016/10/07 04:03:00 ERROR in input [exec]: Errors encountered: [exec: exit status 1 for command '/opt/temp.py']
2016/10/07 04:03:10 Output [influxdb] buffer fullness: 13 / 10000 metrics. Total gathered metrics: 13. Total dropped metrics: 0.
2016/10/07 04:03:12 Output [influxdb] wrote batch of 13 metrics in 12.768427ms
2016/10/07 04:03:20 ERROR in input [exec]: Errors encountered: [exec: exit status 1 for command '/opt/temp.py']
2016/10/07 04:03:20 Output [influxdb] buffer fullness: 18 / 10000 metrics. Total gathered metrics: 31. Total dropped metrics: 0.
2016/10/07 04:03:20 Output [influxdb] wrote batch of 18 metrics in 12.039733ms
2016/10/07 04:03:30 Output [influxdb] buffer fullness: 0 / 10000 metrics. Total gathered metrics: 31. Total dropped metrics: 0.
2016/10/07 04:03:40 ERROR in input [exec]: Errors encountered: [exec: exit status 1 for command '/opt/temp.py']

Additional info:

Permissions
-rwxrwxrwx 1 root root 395 Oct 7 04:05 temp.py

temp.py

Any assistance would be greatly appreciated. I am loving Influxdb and Telegraf!

[sparrc]

Hi @roycepope, I see that you've set the correct permissions to execute the binary, but I wonder if there are still permissions issues with some of the operations in the script.

What happens when you run the script as the telegraf user? try sudo -u telegraf /opt/temp.py

[roycepope]

Thanks @sparrc I think this is definitely a permission issue.

sudo -u telegraf /opt/temp.py

Traceback (most recent call last):
  File "/opt/temp.py", line 12, in <module>
    result = instance.read()
  File "/opt/dht11.py", line 34, in read
    RPi.GPIO.setup(self.__pin, RPi.GPIO.OUT)
RuntimeError: No access to /dev/mem.  Try running as root!

[wangmin19861205]

I also encountered the same problem, how did you solve it?

[roycepope]

@wangmin19861205 I ran the script as root

[forzagreen]

I added telegraf user to sudo and root groups:
sudo usermod -aG sudo telegraf and sudo usermod -aG root telegraf,
and I added telegraf ALL=(ALL) NOPASSWD: ALL to /etc/sudoers,
Then I run my command as sudo.

[alatui]

Adding the user telegraf to the root group is not a good idea.
I came across the same problem when using RaspberryPi with GrovePi, it turns out that I had only to add the user telegraf to the i2c group:

usermod -a -G i2c telegraf

[seans11]

In dealing with the exec plugin it might be useful to grant necessary privileges using file ACLs, for example:

# change owner/group of input script to telegraf, and grant group telegraf read-execute
chown telegraf:telegraf /usr/local/bin/collect_data.py
setfacl -m g:telegraf:rx /usr/local/bin/collect_data.py

In addition consider what privileges your collection script might need. If you are parsing a log file under /var/log you will likely need to do the following:

setfacl -m g:telegraf:r /var/log/my.log

If this log is getting rotated by logrotate process, don't forget to add the ACL grant in the rotate script:

# /etc/logrotate.d/mylog
{
   size 1M
   rotate 10
   create 0640 root root
   sharedscripts
   postrotate
      setfacl -m g:telegraf:r /var/log/my.log
   endscript
}

[IntegersOfK]

I came here from a Google search and fixed a similar error on my Raspberry Pi 4 with sudo usermod -G video telegraf

[luyanan27]

vim /usr/lib/systemd/system/telegraf.service，then change "User=telegraf" to "User=root" can solve the problem

[mageru]

For me the fix was changing /etc/passwd telegraf user shell from /bin/false to /bin/bash