m2r
m2r copied to clipboard
Support rolling restart
rolling "zero-downtime" (asynchronous) reloading strategy similar to Passenger's rolling reloads. This has been the default since 1.1.0 up till the 1.3.x line. If you use this you should account that your JVM memory requirements might increase quite a lot (esp. if you reload under heavy loads) since requests are being served while there's another version of the application being loaded.
https://github.com/trinidad/trinidad#hot-deployment