RATIONALE.md

PANZER

Worker: Monitor starts a worker process:

1. run a Worker instance in its own fork

Monitoring: Panzer starts a monitor process:

1. load application, execute global configuration (e.g. TCPServer.new)
2. when monitor has loaded: send SIGVTALRM to monitor process
3. forks N worker processes
4. on SIGCHLD: 1. collect zombie worker processes 2. fork worker processes to replenish pool
5. on SIGTERM: 1. send SIGTERM to worker processes (exit gracefully) 2. don't replenish pool anymore 3. send SIGINT to worker processes (exit now) after timeout 4. exit
6. on SIGINT (first time): 1. behave as SIGTERM

TODO:

7. on SIGINT (second time): 1. send SIGINT to worker processes (exit now) 2. exit
8. on SIGTTIN: 1. print monitor status 1. print workers' status

Zero downtime: Panzer starts a main process:

1. forks/execs monitor process
2. on SIGUSR1: 1. forks/execs a new monitor process (which then forks new worker processes)
3. on SIGVTALRM: 1. tells old monitor process to gracefully terminate (SIGTERM) 2. tells old monitor process to exit after timeout (SIGINT)
4. on SIGTERM: 1. send SIGTERM to monitor process (exit gracefully) 2. exit
5. on SIGINT (first time): 1. behave as SIGTERM
6. on SIGINT (second time): 1. send SIGINT to monitor process (exit now) 2. exit

TODO:

7. on SIGTTIN: 1. print main process status 2. send SIGTTIN to monitor process

TREE

Running state:

• panzer:main
  • panzer:monitor
    • panzer:worker (1)
    • panzer:worker (2)
    • panzer:worker (3)
    • panzer:worker (4)

Restarting state:

• panzer:main
  • panzer:monitor (exiting)
    • panzer:worker (2, exiting)
    • panzer:worker (4, exiting)
  • panzer:monitor (new)
    • panzer:worker (1)
    • panzer:worker (2)
    • panzer:worker (3)
    • panzer:worker (4)