Adapt Hipache to work with OpenRuko #16

Open
Filirom1 opened this Issue Mar 29, 2013 · 0 comments

Comments

Projects
None yet
1 participant
Owner

Filirom1 commented Mar 29, 2013

This is an old topic:
https://groups.google.com/d/msg/openruko/kSJxjqeawhc/lGJ926XiuyoJ

We have started it :
https://github.com/Filirom1/apiserver/commits/hipache
https://github.com/Filirom1/hipache/commits/openruko

And I dont' know if it's a good idea.

httprouting and Hipache use node-http-proxy opensourced by the PaaS company : nodejitsu.

Hipache depends on Redis, which is a great tool with a small footprint. But it's one more component to install. Heroku/Logplex too depends on Redis: openruko/logplex#4

If a compare Hipache with httprouting :

  • Load-balancing across multiple backends : OK with httprouting
  • dead-backend-detection but now it's easy with node-http-proxy to react on a bad dyno response : nodejitsu/node-http-proxy@3b86a7a. And this job is done by rukorun and the heartbeat
  • Multi-process architecture : done in 5min with cluster-master
  • Memory monitoring : ouch this a nodejs problem or a node-http-proxy problem. BTW cluster-master could restart node gracefully periodically. That solve the problem
  • Dynamic configuration : configuration is read in PostgreSQL instead of Redis
  • WebSocket : done
  • SSL : slow in nodejs openruko/httprouting#9
  • Custom HTML error pages : interesting but would it works in the Heroku way : openruko/httprouting#11
  • Wildcard domains support : Good : openruko/httprouting#10

And I would like to add performance openruko/httprouting#8

Moreover there is a lot of features to be implemented to become a true clone of Heroku, that are not in Hipache. I fear that it will be harder to implement in Hipache than in httprouting.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment