Ping Input Plugin adds a success entry to Influxdb even during failure

[prashanthjbabu]

Relevant telegraf.conf:

[[inputs.ping]]
urls = ["8.8.8.8"] # required

System info:

prash@prash-laptop:/Downloads/telegraf/usr/bin$ ./telegraf -version
Telegraf v1.8.05f8c983 (git: master 5f8c983)

Description:

Telegraf Ping Plugin does not support certain versions of ping . For example , the busybox version of ping is unsupported since it does not support the interval option (-i) which is needed by the ping plugin.In such cases , it is seen that an entry is made into influx as shown below :
`2018-08-13T11:34:10Z E! Error in plugin [inputs.ping]: Fatal error processing ping output: 192.168.1.1

ping,host=3ebf7bd6fcd0,url=192.168.1.1 result_code=0i 1534160050000000000

2018-08-13T11:34:20Z E! Error in plugin [inputs.ping]: Fatal error processing ping output: 192.168.1.1

ping,host=3ebf7bd6fcd0,url=192.168.1.1 result_code=0i 1534160060000000000

2018-08-13T11:34:30Z E! Error in plugin [inputs.ping]: Fatal error processing ping output: 192.168.1.1
`
As seen above , result_code = 0 meaning success(according to the documentation). This can be misleading since it could imply a successful ping even though ping didn't run at all.

Steps to reproduce:

1. Run an alpine docker - docker run -ti alpine sh
2. Copy telegraf and a config file which has ping options
3. Run telegraf and you will see the output above.

Expected behavior:

result_code should indicate an error instead of success.
Example:
ping,host=e3e8019bc54b,url=192.168.1.1 result_code=2i 1534160890000000000
where result_code=2 can indicate ping error

Actual behavior:

result_code output is "0" indicating successful ping.

Additional info:

[prashanthjbabu]

This issue is addressed in PR #4550

[glinton]

Can you paste your full [[inputs.ping]] config please? It shouldn't be setting the interval unless ping_interval is set.

[prashanthjbabu]

Here's the entire telegraf.conf file that I'm using

/ # cat /tmp/telegraf.conf
[[inputs.ping]]
urls = ["www.google.com"]
[[outputs.file]]
files = ["stdout"]

The error thrown is :

/bin/ping: unrecognized option: i
BusyBox v1.28.4 (2018-05-30 10:45:57 UTC) multi-call binary.

Usage: ping [OPTIONS] HOST

Send ICMP ECHO_REQUEST packets to network hosts

-4,-6		Force IP or IPv6 name resolution
-c CNT		Send only CNT pings
-s SIZE		Send SIZE data bytes in packets (default 56)
-t TTL		Set TTL
-I IFACE/IP	Source interface or IP address
-W SEC		Seconds to wait for the first response (default 10)
		(after all -c CNT packets are sent)
-w SEC		Seconds until ping exits (default:infinite)
		(can exit earlier with -c CNT)
-q		Quiet, only display output at start
		and when finished
-p HEXBYTE	Pattern to use for payload

Looking at the code , it is quite evident that pingInterval has a default value

return &Ping{
pingHost: hostPinger,
PingInterval: 1.0,
Count: 1,
Timeout: 1.0,
Deadline: 10,
}

And the -i option gets set when the PingInterval is > 0

if p.PingInterval > 0 {
args = append(args, "-i", strconv.FormatFloat(p.PingInterval, 'f', -1, 64))
}

I tried changing the default PingInterval value to 0.0 and it worked , although I'm not sure if there would be any collateral by doing so.

[glinton]

You are correct, I only read the Readme which says it defaults to 0. Setting it to 0 will be fine for your case. I'm going to leave this open and rename it as an issue with the ping readme.

[prashanthjbabu]

So just to make it clear , I just need to set "ping_interval" to 0.0 explicity in telegraf.conf file and it would work for busybox and other such ping applications which don't have too many ping options(like -i for interval).

[glinton]

correct, but there is still the issue as you originally described it, where a valid ping metric gets recorded by telegraf rather than it failing on error.

[prashanthjbabu]

Yes that's correct ! The PR mentioned earlier addresses that! Once that goes in , it should be resolved!

[danielnelson]

Closed in #4550