Telegraf package for pfSense 2.3.5 and 2.4.0 snapshots is now available.

overcookedTOFU

I think the contents of the password field are not being passed properly to the telegraf.conf file.

Specifically, the password is being encoded to sanitize inputs, but is not being decoded before the characters are written to the config file. See line 78 in telegraf.xml.

The password in the telegraf.conf file ends up being the base64 encoding of the password entered in the password field. With the wrong password in the configuration file, telegraf fails to connect to the influxdb database.

One can work around the error by manually editing the telegraf.conf file located at /usr/local/etc/telegraf.conf to correct the password.

Note: By default, influxdb does not require username/password authentication. If authentication is not enabled on the influxdb server, then (as far as I know) telegraf will not complain about an incorrect password.

fbearoff

@overcookedTOFU:

I think the contents of the password field are not being passed properly to the telegraf.conf file.

Specifically, the password is being encoded to sanitize inputs, but is not being decoded before the characters are written to the config file. See line 78 in telegraf.xml.

The password in the telegraf.conf file ends up being the base64 encoding of the password entered in the password field. With the wrong password in the configuration file, telegraf fails to connect to the influxdb database.

One can work around the error by manually editing the telegraf.conf file located at /usr/local/etc/telegraf.conf to correct the password.

Note: By default, influxdb does not require username/password authentication. If authentication is not enabled on the influxdb server, then (as far as I know) telegraf will not complain about an incorrect password.

I am experiencing the same issue, the base64 encoding ends up in the password field of the telegraf.conf file. This does not persist through an upgrade either, I have to reset the file each time.

ivor

Please submit a bug on https://redmine.pfsense.org/

lordbob75

Hopefully this isn't a stupid question: How do I install this? I've tried a couple different ways but it can never find the package.

I am running an SG-1000 if that matters.

ivor

@lordbob75:

Hopefully this isn't a stupid question: How do I install this? I've tried a couple different ways but it can never find the package.

I am running an SG-1000 if that matters.

System > Packages Manager > Available Packages.

If you still cannot see it, update your pfSense version to the latest.

lordbob75

@ivor:

@lordbob75:

Hopefully this isn't a stupid question: How do I install this? I've tried a couple different ways but it can never find the package.

I am running an SG-1000 if that matters.

System > Packages Manager > Available Packages.

If you still cannot see it, update your pfSense version to the latest.

Aha, thought I was already on it.

I just updated however, and I still don't see it listed.

jimp

It's also amd64 only. If you're on 2.3.5 i386 or 2.4 on an SG-1000 you won't see it.

It's disabled on SG-1000 because it requires lang/go14 which does not build on ARM currently.

lordbob75

@jimp:

It's also amd64 only. If you're on 2.3.5 i386 or 2.4 on an SG-1000 you won't see it.

It's disabled on SG-1000 because it requires lang/go14 which does not build on ARM currently.

Ah, of course! :'(

Any idea if that's something that will work in the future? Or is it never gonna happen?

I've been trying to figure out how to get data out to my monitoring database, but haven't managed to get anything working yet and this looked like the best option.

fastjack

Is there a way for Telegraf to deliver metrics about gateway latencies and packet loss?

I know that most of the metrics that Telegraf does deliver at the moment are just the built in input filters. I assume this needs some custom input filter that parses the output of dpinger, though I am not sure how to go about it.

fastjack

@fastjack:

Is there a way for Telegraf to deliver metrics about gateway latencies and packet loss?

Replying to myself here. I did a quick&dirty python script (not my native programming language) that generates metrics from dpinger stats. You can find it here: https://gist.github.com/fastjack/a0feb792a9655da7aa3e2a7a1d9f812f

Just drop it somewhere like /usr/local/libexec/telegraf/ (might have to create that directory) and make it executable.

Then add these lines to your telegraf.conf:

[[inputs.exec]]
        commands = ["/usr/local/libexec/telegraf/gateways.py"]
        timeout = "5s"
        data_format = "influx"

Adjust the path in commands if you copied the script somewhere else. Don't forget to restart Telegraf afterwards!

fastjack

While playing around with Telegraf I also noticed this in /var/log/telegraf.log:

2017-08-30T07:56:40Z E! Error writing to output [influxdb]: Could not write to any InfluxDB server in cluster

The reason seems to be a bug during the generation of telegraf.conf. Besides the real [[outputs.influxdb]] section there is a second empty [[outputs.influxdb]] section generated. Once I removed that empty section the error messages disappeared.

doktornotor

https://github.com/pfsense/FreeBSD-ports/pull/410