-
Notifications
You must be signed in to change notification settings - Fork 909
Description
When a SIGHUP is sent to a container (ie. using something like docker kill -s HUP container-ID), the container is deregistered, but never reregistered, even though the container is still running.
My use case is with Nginx and docker-gen running in separate containers. I'm using an Nginx container to proxy requests to one or more API servers running in containers, and I'm using docker-gen to monitor for new API server containers in order to update and reload (via SIGHUP) the Nginx config. See here for more details regarding this use case.
I'm using registrator to make my Nginx proxy container available via my Consul-based DNS server, and find that when new API server containers are added the DNS entry for the Nginx proxy goes away. I can restart the Nginx proxy via docker restart and it shows back up again.
I suppose the feature requested in #128 could be a fix for this, but am wondering if the SIGHUP event can be handled directly instead.