Error: Could not write to any InfluxDB server in cluster

[daniel-alonso-sanchez]

Bug report

Hi. First of all, thanks a lot for your support and the tool!
I've a problem with latest telegraf versions (1.4.2 and 1.4.3), tryining to write into influxdb (1.3.6).
I've tested also with a previous version (telegraf 1.2.1) in windows and linux machines and it works perfectly

Relevant telegraf.conf:

[agent]

  interval = "10s"

  round_interval = true


  metric_buffer_limit = 1000

  flush_buffer_when_full = true


  collection_jitter = "0s"


  flush_interval = "10s"

  flush_jitter = "0s"
[[outputs.influxdb]]

  urls = ["http://10.19.24.123:8086"] 

  database = "telegraf" # required

  precision = "s"


  timeout = "5s"

System info:

• Suse Linux 12
• Windows 8

Telegraf 1.4.3 and 1.4.2

Steps to reproduce:

1. Start a clean installation of influxdb
2. Configure and start telegraf

Expected behavior:

Metrics stored into influxdb

Actual behavior:

2017-10-26T11:52:10Z E! InfluxDB Output Error: Response Error: Status Code [403]
, expected [204], [<nil>]
2017-10-26T11:52:10Z E! Error writing to output [influxdb]: Could not write to a
ny InfluxDB server in cluster

Additional info:

[Include gist of relevant config, logs, etc.]

There's a corporate proxy, but everything works with telegraf 1.2.1

Use case:

POC in order to show to managers and OPS team the advantages of gathering metrics and alerting using the stack (chronograf, telegraf , kapacitor)

Thanks again!

[danielnelson]

403 Forbidden usually indicates that the user does not have write permissions to the database, but in this case I think this may be coming from the proxy. Normally a permission error from InfluxDB has an error message like:

E! InfluxDB Output Error: Response Error: Status Code [403], expected [204], ["howdy" user is not authorized to write to database "telegraf"]

[daniel-alonso-sanchez]

You're porobably right. Because also, there's no log entry un influxdb about the problem... But i cannot understand why everything works with 1.2.1, using same config files...

I'm gonna try sniffing net packsges...

[daniel-alonso-sanchez]

Confirmed. It's the annoying corporate proxy. It's funny because it's the first company that limits also connections among internal machines...

Is there a way of setting a NO_PROXY exception in source machine (telegraf machine)?

Thanks a lot and sorry for the inconvenience.

[daniel-alonso-sanchez]

Looks like there is other people asking for something similar: #1588

Closing this ticket. Thanks again!

[danielnelson]

That makes sense, we only added support for the proxy environment variables recently.

It sounds like maybe we need a way to disable this on a per plugin basis in the config, but in the meantime you can add NO_PROXY="*" in /etc/default/telegraf.

[daniel-alonso-sanchez]

Hi again! That's exactly what i've done and works like a charm! Thanx a lot!!

[senjirui]

Telegraf (version 1.3.5) I met the same problem. I have done like this ==>add NO_PROXY="*" in /etc/default/telegraf . But,it does not fix it

[danielnelson]

@senjirui We only added proxy envvar support in 1.4.2, so I think you must have a different issue.