1 |
On Fri, Aug 17, 2012 at 3:29 AM, Luca Barbato <lu_zero@g.o> wrote: |
2 |
> On 8/17/12 7:39 AM, heroxbd@×××××.com wrote: |
3 |
>> Thanks a lot for the comment. I agree. Nobody cares about the tiny (not |
4 |
>> tested) booting up time gain with rc_parallel. |
5 |
> |
6 |
> It can be not so tiny, surely busybox+openrc gives a better gain in many |
7 |
> cases. |
8 |
> |
9 |
|
10 |
I suspect that it will depend greatly on what services you're running, |
11 |
and what order they happen to start in, and what you care about. In |
12 |
theory slamming the kernel with a ton of processes will allow it to |
13 |
manage its queues better with a fuller understanding of demand. |
14 |
systemd can potentially short-cut this a bit further since it can |
15 |
consider a dependency resolved if nothing more than a socket is |
16 |
created, which is a clever trick (I have no idea how well it works out |
17 |
in practice, though I have used a .socket service once and that worked |
18 |
out fine (with the caveat that the first connection fails)). |
19 |
|
20 |
Where I saw the bigger performance difference between openrc and |
21 |
systemd was in shutdown. Systemd shuts down REALLY fast, and I've |
22 |
noticed it tends to actually kill ssh sessions rather than leaving |
23 |
them unresponsive (plus ssh dies even faster so that makes it seem |
24 |
subjectively faster - but the difference in full shutdown is still |
25 |
real). Obviously people don't care as much about shutdown |
26 |
performance, but on a laptop (where supposedly event-driven inits |
27 |
should shine) I know that waiting for a full shutdown before packing |
28 |
up can be painful at least in the land of Windows. |
29 |
|
30 |
Just food for thought - it seems odd that parallel startup should have |
31 |
any issues at all, provided that service scripts don't terminate until |
32 |
the service is functional and all dependencies are specified. If |
33 |
either of those aren't true then you'll have race conditions. |
34 |
|
35 |
Rich |