We have had an amount of servers in AWS, successfully running rails3 application on ubuntu10. We use capistrano for deploy. Servers are behind the load balancer, running unicorns behind nginx.
One day we updated the configuration and turned on two new servers running ubuntu14. We deployed the app onto them in staging mode, made sure everything goes well, and re-deployed the app there as production.
After new servers were turned on, we started to encounter problems on some requests with assets pipeline: not all assets were available. I suspect there was kinda désynchronisation between old good ubuntu10s and new ubuntu14. The reproducible behaviour is:
- 14 on ⇒ some assets fail;
- 14 off ⇒ everything is great.
We tried to clean / recompile assets manually, with no luck. I believe there is something really easy I am missing. So, my question would be: how to properly introduce new servers, running Rails3 app, in Amazon farm, so that assets are not messed up?
Aucun commentaire:
Enregistrer un commentaire