Hi Paul.
Am 14-11-2015 14:00, schrieb Paul Menzel:
Dear HAProxy folks,
I am using a Docker setup to serve Web application using a database.
HAProxy, the Web app and the database each run in a separate Docker
container.
The HAProxy container is than started with the command below.
docker run --detach --name haproxy --link webapp:webapp
myimages/haproxy
The goal is to have zero downtime deployments.
Now the problem is, that each time the Web app container is replaced by
a container run from a newer version, the HAProxy container has to be
stopped an started again so that the link works. This is certainly not
desired.
I search the Web for documentation and how-tos but couldn’t find
anything directly related to my issue. Do you know of such write-ups,
which I might have missed due to the generic key words.
Do you have any experiences with serving Web apps with HAProxy using
Docker links?
Should I stop using that “technology” and do things differently?
You can take a look into kuberbetes, openshift or similar frameworks for
docker management.
http://getcloudify.org/2015/06/11/orchestration-docker-cloud-automation-openstack-heat-tosca-kubernetes.html
https://github.com/kubernetes/kubernetes
http://kubernetes.io/v1.1/docs/user-guide/replication-controller.html#rolling-updates
https://github.com/openshift/origin
https://docs.openshift.org/latest/dev_guide/deployments.html
Due to the fact that this frameworks use docker inside you can reuse
some of the dockerfiles.
I use this openshift stuff with rolling updates.
BR Aleks
Thanks,
Paul
[1]
http://docs.docker.com/engine/userguide/networking/default_network/dockerlinks/
