[
https://issues.apache.org/jira/browse/AURORA-761?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14368806#comment-14368806
]
Stephan Erb commented on AURORA-761:
------------------------------------
FYI, we ended up doing the following:
* Deploy the patched synapse gem by [~benley] to each slave (thanks!)
* Synapse requires us to maintain a list of aurora services that shall be
exposed via fixed HAProxy endpoints. The services deployed in our cluster are
highly dynamic. We therefore opted to start one synapse/haproxy pair per aurora
service within the same container as the service itself. The synapse config is
generated to contain all external services needed by the primary service in the
container.
* Expose the HAProxy endpoints to the primary service using environment
variables
* Rewire the incoming health checks to /health to pass through haproxy to make
sure it remains operational (as it is running in the background and cannot be
monitored by thermos)
> Provide a proxy for generic service discovery
> ---------------------------------------------
>
> Key: AURORA-761
> URL: https://issues.apache.org/jira/browse/AURORA-761
> Project: Aurora
> Issue Type: Story
> Components: Service Discovery, Usability
> Reporter: Bill Farner
> Priority: Minor
>
> While {{Announcer}} provides service registration, we lack a cross-cutting
> answer for service discovery. There are well-known libraries that will do it
> (e.g. finagle), but we need an answer for others. Marathon, for example,
> provides a script called {{haproxy_marathon_bridge}} that reloads
> configuration of HAProxy for this purpose. We could do something similar
> with a mixin {{Process}} that dynamically routes an inbound port to a
> serverset path in ZooKeeper.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
