TODO 3.7 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970
  1. For version 0.17:
  2. -----------------
  3. * Limit memory use by control connections. Currently clients have command responses queued without
  4. limit; it would be better to stop accepting new commands once a certain amount of response is
  5. buffered.
  6. * Externally triggered services (to represent devices etc). Such services will not reach STARTED
  7. state unless the external trigger occurs (eg "dinitctl trigger servicename", details to be
  8. determined).
  9. For version 1.0 (release requirements):
  10. ---------------------------------------
  11. * Be able to boot and shutdown Linux and FreeBSD (or OpenBSD).
  12. * Documentation must be complete (mostly done; will need updates as other items are completed).
  13. Maybe for 1.0?
  14. --------------
  15. * on shutdown, after a long interval with no activity, display information about
  16. services we are waiting on (and/or, do this when prompted via ^C or C-A-D?)
  17. * Proper support for socket activation?
  18. * Chaining of service process input/output?
  19. For later (post 1.0):
  20. ---------------------
  21. * jails support
  22. * On linux when running with PID != 1, write PID to /proc/sys/kernel/cad_pid so
  23. that we still receive SIGINT from ctrl+alt+del (must be done after /proc is
  24. mounted, possibly could be left to a service script)
  25. * Perhaps need a way to prevent script services from re-starting.
  26. (eg there's no need to mount filesystems twice; there might be various other
  27. system initialisations that can't or shouldn't really be "undone" and so do
  28. not need to be re-done).
  29. * Internationalisation
  30. * A service can prevent shutdown/reboot by failing to stop. Maybe make
  31. multiple CTRL-ALT-DEL presses (or ^C since that's more portable) commence
  32. immediate shutdown (or launch a simple control interface).
  33. * When we take down a service or tty session, it would be ideal if we could kill
  34. the whole process tree, not just the leader process (need cgroups or pid
  35. namespace or other mechanism).
  36. * Allow logging tasks to memory (growing or circular buffer) and later
  37. switching to disk logging (allows for filesystem mounted readonly on boot).
  38. But perhaps this really the responsibility of another daemon.
  39. * Allow running services with different resource limits, chroot, cgroups,
  40. namespaces (pid/fs/uid), etc
  41. * Support chaining service output to another process (logger) input; if the
  42. service dies the file descriptor of its stdout isn't closed and is reassigned
  43. when the service is restarted, so that minimal output is lost.
  44. - even more, it would be nice if a single logger process could be responsible
  45. for receiving output from multiple services. This would require some kind of
  46. protocol for passing new output descriptors to the logger (for when a
  47. service starts).
  48. * dinitcheck should be able to resolve variable substitutions using variables from running dinit.
  49. Even later / Maybe never:
  50. -------------------------
  51. * Support recognising /etc/init.d services automatically (as script services, with
  52. no dependency management - or upstart compatible dependency management)
  53. Also BSD's rc.d style scripts (PROVIDE, REQUIRE).
  54. * Place some reasonable, soft limit on the number of services to be started
  55. simultaneously, to prevent thrashing. Services that are taking a long time
  56. to start don't count to the limit. Maybe use CPU/IO usage as a controlling
  57. factor.
  58. * Cron-like tasks (if started, they run a sub-task periodically. Stopping the
  59. task will wait until the sub-task is complete).
  60. * Allow to run services attached to virtual tty, allow connection to that tty (ala "screen").
  61. * SystemD-like handling of filesystem mounts (see autofs documentation in kernel)
  62. i.e. a mount point gets an autofs attached, and lazily gets mounted when accessed
  63. (or is mounted in parallel). Probably put the functionality in a separate daemon.