Open dagendresen opened 6 years ago
I had an Apache httpd on port 80, apachectl stop
resulted in:
(...) ERROR: for dnsmasq Cannot start service dnsmasq: driver failed programming external connectivity on endpoint dnsmasq (f0b86b0407750031fdc829d56769d509dfce11f84ddef3cf8122ff93562a8981): Error starting userland proxy: Bind for 172.17.0.1:53: unexpected error (Failure EADDRNOTAVAIL)
The host running the docker daemon will need to have the 80 and 443 ports "free" if a proxy in a docker-compose.yml file binds to those ports when a composition is started (with docker-compose up -d).
Also if you have many different projects on the same host that you run simultaneously, only one proxy can be running - for the same reason - it is a singleton in that sense. It is possible to "break out" the singletons into a separate project - in this case it would be dnsmasq and proxy services - then you could run any number of different projects simultaneously if removing these services from those projects and making sure the same software defined networks are used across projects/compositions.
There are some important post-install steps required after installing docker on the host.
If the user running the docker daemon is not in the docker group various issues can pop up related to permissions on the system. There shouldn't ever be the need to use "sudo" - I think - with the docker command if properly setup. For example, see the the "usermod" step which does adds the user to the docker group here:
bash-script with docker and docker-compose installation steps
To update the status on this issue - the first error seem to come from another service running locally binding to the port (so the resolution is known), the second error appears to be similar - the docker process is not allowed to bind to port 53 on the host. Looks like a permissions issue. If the user running the docker process belongs to the "docker group" this shouldn't be an issue.
So @dagendresen pls confirm that the "post install steps" - which involves adding the local docker user to the docker group - is completed. If it is then this issue is due to something else and further details would be needed, such as logs etc.
Otherwise it could be marked as fixed, pending confirmation from @dagendresen.