Websocket input doesn't work

[jordansissel]

(This issue was originally filed by @aemxdp at https://github.com/elastic/logstash-contrib/issues/161)

---

Config:

input {
  websocket {
    url => "ws://127.0.0.1:3000"
  }
}

output {
  stdout {}
}

Run:

$ /opt/logstash/bin/logstash agent -f logstash-simple.conf

Output:

Using milestone 1 input plugin 'websocket'. This plugin should work, but would benefit from use by folks like you. Please let us know if you find bugs or have suggestions on how to improve this plugin.  For more information on plugin milestones, see http://logstash.net/docs/1.4.2-modified/plugin-milestones {:level=>:warn}
websocket input client threw exception, restarting {:exception=>#<NoMethodError: undefined method `redirect?' for nil:NilClass>, :level=>:warn}
websocket input client threw exception, restarting {:exception=>#<NoMethodError: undefined method `redirect?' for nil:NilClass>, :level=>:warn}
websocket input client threw exception, restarting {:exception=>#<NoMethodError: undefined method `redirect?' for nil:NilClass>, :level=>:warn}
...

Websocket example (nodejs/coffee):

#!/bin/env node
WebSocket = require "ws"
wss = new WebSocket.Server(port: 3000)
foo = -> wss.clients.forEach (client) -> client.send("test")
setInterval(foo, 3000)

I've checked it works by writing test client like this:

WebSocket = require "ws"
ws = new WebSocket("ws://127.0.0.1:3000")
ws.on "message", (data) -> console.log(data)

[bradvido]

Using logstash 1.5.0 it is still failing:
:message=>"websocket input client threw exception, restarting", :exception=>#<NoMethodError: undefined method each' for #FTW::Response:0x24b988a3>, :level=>:warn}`

---

Actually further investigation shows that what is actually happening is that the websocket failed to execute the handshake so the FTW agent returned the http response instead of the websocket. Since the response has no each method, we get an exception.

It would be better to check if the returned object is a response and throw an exception saying the websocket failed to handshake (or if the underlying library threw the exception itself)

[magnusbaeck]

@bradvido or @aemxdp, have you tried adding a trailing slash to the URL (i.e. ws://127.0.0.1:3000/)? The upstream ruby-ftw library can't deal with slashless URLs (see jordansissel/ruby-ftw#29) and doesn't recover gracefully when the Node.JS peer terminates the connection open receiving a malformed request from ruby-ftw.

[PaulCheriyan009]

I have facing same issue. using node server and added trailing slash, but still not working

input { websocket { mode => "client" url => "wss://localhost:9300/" type => "string" } } output { stdout { } }

[GitArunsh]

i had same issue, adding a trailing slash to url in logstash pipeline solved the issue. here is the logstash config file
input {
websocket {
mode => "client"
url => "ws://127.0.0.1:8080/"
type => "string"
}
}
output {
stdout { codec => rubydebug }
}

[wodphoenix]

I have same error. My config

input {
websocket {
url => "ws://10.0.30.54:200/"
codec => "json"
mode => "client"
type => "string"
}
}

output {
elasticsearch {
hosts => ["http://localhost:9200"]
index => "log-%{+YYYY.MM.dd}"
}
}

[billndotnet]

Maybe related, but how does this plugin handle authentication requests for WSS endpoints, like Oauth? It doesn't look like it has any support for that, from the docs?

[p4ulmac]

I'm seeing the same issue, and have tried various iterations of the connection string, wss, trailing slash - no trailing slash, but just can't get it to work. The JSON and websocket has all been verified using various tools, so its at the point now when everything points to the logstash plugin. Happy to help with testing if there are any updates made, just don't have quite enough knowledge to have a stab at fixing it!

Whenever I enable my pipeline for the websocket input, this is thrown in the log;
websocket input client threw exception, restarting {:exception=>#<HTTP::Parser::Error: invalid method>