v1.6.7

• Fixed the issue that there are multiple haproxy process running, causing 503 errors

v1.6.6

• Updated the python SDK version which has a 15 seconds timeout for dockercloud API calls
• Added a set of EXTRA_XXX envvars to allow haproxy load from files
• Fixed a nonetype error when getting a label from a task
• Allowed ADDITIONAL_BACKEND_ containing more that 9 chars
• Added a default frontend if the EXTRA_FRONTEND_SETTINGS is used

v1.6.5

• Add Reload time to force reload haproxy given a timeout

v1.6.4

• Fix the wrong description of ADDITIONAL_BEACKENDS in README
• ADD HAPROXY_USER and HAPROXY_GROUP envvar to make the user and group configurable when when binding a unix socket.

v1.6.3

• add an option EXCLUDE_BASIC_AUTH to disable basic auth in a selected service
• add an option FAILOVER to allow a service to be configured as a failover to another service
• add an option HTTP_BASIC_AUTH_SECURE to set hashed credentials instead of plain text
• remove DHE-RSA-AES128-SHA from default SSL_BIND_CIPHERS
• do not use container ip address attached in the ingress network

v1.6.2

• add support turn on/off a service from adding to haproxy, which is useful for blue/green testing.

e.g. Imagine you have a service called blue, and another called green and both are load balanced by dockercloud haproxy. You can do the following to remove the green service,

docker service update --label-add "com.docker.dockercloud.haproxy.deactivate=true" green

and run the following command to add it back

docker service update --label-add "com.docker.dockercloud.haproxy.deactivate=false" green

v1.6.1

• update the sdk version, fixing the issue that 401 error never get retried correctly

v1.6.0

• allow adding additional backends thru envvar(see: #99)
• do not re-config haproxy if the related events are accumulated
• Restart haproxy if it dies unexpectedly
• support docker swarm mode (experimental)

v1.5.3

update base image from alpine:edge to alpine:3.4

v1.5.2

• Fix the error when /reload.sh is invoked