dgoss add port in tomcat images shows not listening even though the port is listening

[frezbo]

docker pull tomcat:alpine
dgoss edit -p 8080:8080 tomcat:alpine

INFO: Starting docker container
INFO: Container ID: de8ef54c
INFO: Run goss add/autoadd to add resources
/goss # goss add port 8080
Adding Port to './goss.yaml':

tcp:8080:
  listening: false
  ip: []


/goss # netstat -tlnp
Active Internet connections (only servers)
Proto Recv-Q Send-Q Local Address           Foreign Address         State       PID/Program name    
tcp        0      0 :::8080                 :::*                    LISTEN      1/java
tcp        0      0 ::ffff:127.0.0.1:8005   :::*                    LISTEN      1/java
tcp        0      0 :::8009                 :::*                    LISTEN      1/java
/goss # wget http://localhost:8080
Connecting to localhost:8080 (127.0.0.1:8080)
index.html           100% |*******************************| 11230   0:00:00 ETA
/goss # 

[aelsabbahy]

This is the same issue described in #149, it should work fine if you do goss add port tcp6:8080

[frezbo]

@aelsabbahy thanks, but will this be fixed?

[aelsabbahy]

I definitely want it fixed, but haven't spent the time trying figure out the best way to do this.

netstat doesn't handle it, but ss does. So I probably need to dig into the ss code to see how it handles this.

[aelsabbahy]

Hmm, just reread your netstat output. Wonder if it was an older version of netstat that would show tcp6.

[frezbo]

Maybe as I am using alpine I am not able to see the version. This is the busybox version

BusyBox v1.25.1 (2016-10-26 16:15:20 GMT) multi-call binary.

[pysysops]

I don't mind having a look at this one (as there's a few open issues for it) before I re-visit the HTTP headers / gomega attempt.

[aelsabbahy]

Sure. For what it's worth, netstat on busybox looks like a dead end.

Think it prints "tcp" for both tcp and tcp6:
http://code.metager.de/source/xref/busybox/networking/netstat.c

Sort of, kind of, maybe related (Or not, but interesting read):
golang/go#1854

I think ss handles this correctly, but I haven't verified it or figured out why.

More interesting read: https://tools.ietf.org/html/rfc3493#section-3.7

[aelsabbahy]

I might be over complicating this.. maybe there's a simple solution.. I'm not as familiar with the subject, so take everything I'm saying as assumptions :)

Thanks btw. Also, there's another PR from a contributor that might do the header gomega marcher (it's a marchers for sub-maps).. I'll keep you updated.

[aelsabbahy]

Random thought.. maybe it would be simpler to get rid of the whole tcp: vs tcp6: thing all together.

We can collapse the tests into two categories: tcp (default) and udp, if the user wants to make more specific assertions, they can use the ip: attribute.

The only downside is if a user wants to test one or the other, it would be an awkward syntax:

port:
  tcp:22:
    listening: true
    ip:
      and:
      - not:
          contain-element: '::'
      - contain-element: 0.0.0.0

Thoughts on this approach? I'm just thinking out-loud on this.. not 100% sure it's a good idea yet.

[frezbo]

Thanks

[frezbo]

@aelsabbahy the above code might hinder with the simplicity of the goss syntax, just a suggestion

[aelsabbahy]

I'm gonna close this as a duplicate of #149 so we have one location to track this