2

I edited /etc/init.d/rc on my Debian machine and changed startup boot to startup boot &. Boot time decreases ridiculously, and no race conditions happen at all. Way faster than even Systemd (which for some reason gives me 4 or 5 watts more power consumption, leading to drastically decreased battery life - still don't know why)

Why does this even work though? Don't init scripts tend to have dependencies on each other? My computer has a really fast SSD for everything so each script absolutely does not have time to finish before the next one starts. The only complaint I have is that the machine goes to single-user mode (init 1) rather than rebooting when I press reboot in GNOME. Other than that it seems like magic just happened.

5
  • systemd on debian just calls out to a bunch of serial legacy init scripts anyway - it's not a great distribution for it
    – strcat
    Jun 2, 2013 at 3:23
  • I don't really think so. It does speed up boot a lot but often causes bugs (on my old computer it fails to load VBox modules and Bumblebee, on my new one is somehow fails to turn on rc6 for sandy bridge, causing big power consumption)
    – ithisa
    Jun 2, 2013 at 13:59
  • it speeds up your boot but it's calling out to legacy init scripts to run the services, and it has to assume those are serial - the other problems are again just debian packaging issues too
    – strcat
    Jun 2, 2013 at 18:29
  • In, say, Arch, service stoppers/runners are hard-coded in C?
    – ithisa
    Jun 2, 2013 at 20:39
  • They're unit files starting the service rather than legacy init scripts that are treated as serial, and many use socket activation so the start-up is non-blocking. With unit files systemd knows the dependencies so for things that aren't just socket activated, it can still make them start asynchronously.
    – strcat
    Jun 3, 2013 at 1:56

0

You must log in to answer this question.

Browse other questions tagged .