Port-amiga archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: 7.0 slow multiuser boot



frank%phoenix.owl.de@localhost (Frank Wille) writes:

>Sure. Here are the numbers for a 6.1.5 kernel running 7.99.4 userland:

>[/etc/rc starting at Fri Feb  6 16:17:21 CET 2015]
>[/etc/rc finished at Fri Feb  6 16:22:18 CET 2015]


>There is something which I noticed during the last tests, which might be
>an important hint:

>When you press CTRL-t during multiuser boot from time to time, then you
>see the currently executed command. With 6.1.5 userland everything is
>normal. But with 7.99.4 (and most likely 7.0 as well) you see very frequently
>"cmd: sleep [nanosleep]" !

>Seems that it is now sleeping half of the boot process. But why?


/etc/rc spawns a background shell that sends a "nop" every 3 seconds
and that runs 'sleep 3' for waiting.

Apparently it is supposed to detach from console so that SIGINFO will
not reach it, but apparently that doesn't work.
 
The background shell seems to have no influence on startup speed.

I just tried again with a slightly more modernized system:
 
[/etc/rc starting at Fri Feb  6 20:23:39 GMT 2015]
[/etc/rc finished at Fri Feb  6 21:25:33 CET 2015]

That's now 7.99.4 from Jan 26, still with ntpdate, sshd startup
(and postfix by default).

Compared to the previous situation, ntpdate no longer sets the time
because the "hardware clock" starts with the correct time.

But the wall clock time required for rc processing is down from 
04:30 to 01:54.

-- 
-- 
                                Michael van Elst
Internet: mlelstv%serpens.de@localhost
                                "A potential Snark may lurk in every tree."


Home | Main Index | Thread Index | Old Index