NAT tracker seems to be exiting early...

[tomwilkie]

root     15243  1.2  0.0    760   228 ?        Ss   09:03   0:00  \_ /sbin/runsvdir /etc/service
root     15266  0.0  0.0    736     4 ?        S    09:03   0:00      \_ runsv probe
root     15268 20.2  5.3 381144 54468 ?        Sl   09:03   0:02      |   \_ /home/weave/scope-probe -weave.router.addr
root     15324  0.0  0.0   7076   240 ?        S    09:03   0:00      |       \_ conntrack -E -o xml -p tcp
root     17258  0.0  0.0   1664   616 ?        S    09:04   0:00      |       \_ /bin/sh /usr/bin/weave --local ps
root     17363  0.0  0.0   1664   220 ?        S    09:04   0:00      |           \_ /bin/sh /usr/bin/weave --local ps
root     17364  0.0  0.4 246332  4700 ?        Rl   09:04   0:00      |               \_ docker inspect --format={{.Sta
root     15267  0.0  0.0    736     4 ?        S    09:03   0:00      \_ runsv app
root     15269  0.3  0.6  14800  6244 ?        Sl   09:03   0:00          \_ /home/weave/scope-app

[tomwilkie]

We should make these things restart if they fail.

[tomwilkie]

vagrant@ubuntu-14:~/src/github.com/weaveworks/scope$ docker logs weavescope
<app> 2015/09/22 09:03:57 app starting, version f081a70, ID 706f2b94273507e9
<app> 2015/09/22 09:03:57 listening on :4040
<probe> 2015/09/22 09:03:58 probe starting, version f081a70, ID ubuntu-14
<probe> 2015/09/22 09:03:58 publishing to: localhost:4040, scope.weave.local:4040
<probe> 2015/09/22 09:03:58 conntrack existingConnections error: EOF

[tomwilkie]

This seems to happen as conntrack doesn't output any xml when there are no connections:

vagrant@ubuntu-14:~/src/github.com/weaveworks/scope$ sudo conntrack -L -o xml -p tcp --any-nat 2>/dev/null
vagrant@ubuntu-14:~/src/github.com/weaveworks/scope$ 

Whereas I was expecting at least a empty xml response.