1 |
Tom Wijsman posted on Sun, 16 Jun 2013 20:23:24 +0200 as excerpted: |
2 |
|
3 |
> On Sun, 16 Jun 2013 19:21:38 +0200 Pacho Ramos <pacho@g.o> wrote: |
4 |
> |
5 |
>> El dom, 16-06-2013 a las 10:09 -0700, Brian Dolbec escribió: |
6 |
>> [...] |
7 |
>> > Thank you for considering helping. I have stayed away form the |
8 |
>> > intricate details of package management in the past, but I also do |
9 |
>> > not like how long portage is taking now for dep calculations. |
10 |
>> |
11 |
>> And, cannot that efforts be put in enhancing portage instead? |
12 |
> |
13 |
> To make you see the problems and decisions, I'm going to elaborate a |
14 |
> little and would like you to ask yourself some questions. |
15 |
> |
16 |
> Is it possible to reasonable enhance the Portage code to improve dep |
17 |
> calculations in a reasonable amount of time? |
18 |
|
19 |
TL;DR: SSDs help. =:^) |
20 |
|
21 |
Quite apart from the theory and question of making the existing code |
22 |
faster vs. a new from-scratch implementation, there's the practical |
23 |
question of what options one can actually use to deal with the problem |
24 |
/now/. |
25 |
|
26 |
FWIW, one solution (particularly for folks who don't claim to have |
27 |
reasonable coding skills and thus have limited options in that regard) is |
28 |
to throw hardware at the problem. |
29 |
|
30 |
I recently upgraded my main system to SDD. As it happens, my primary |
31 |
boot didn't speed up much[1], but having both the main system partition |
32 |
(bindirs/libdirs/etc) and the portage tree and overlays on SSD |
33 |
*DRAMATICALLY* improved portage's update-ask-deep-newuse-@world |
34 |
dependency resolution time, both for the cold-tree-cache case, and, |
35 |
surprisingly, even (apparently, I've not timed it but I was sometimes |
36 |
annoyed by the time before especially for hot-cache case, and it hasn't |
37 |
bothered me at all since the upgrade) for the hot-cache case. |
38 |
|
39 |
Between that and the 6-core bulldozer[3] I upgraded to last year, I'm |
40 |
quite happy with portage's current performance, even doing routine |
41 |
rebuilds of the perhaps half of kde I have installed, plus some other |
42 |
packages with @live-rebuild.[2] |
43 |
|
44 |
The SSD doesn't have to be particularly big, either, but fast (if you're |
45 |
running SATA3 to use it) is nice. I figured ~64 gig usage here, tho I |
46 |
wanted some overprovisioning, so thought I'd do 128 gig or so. I ended |
47 |
up with 256 gig, only ~60% partitioned (130-some gig) even with |
48 |
duplicate backup partitions for everything. System, tree, /home, etc, on |
49 |
SSD, but still doing spinning rust for my media partitions and third-copy |
50 |
(second backup) partitions, on demonstrated reliable here reiserfs, while |
51 |
the SSD is all still-development-so-keep-your-backups-updated btrfs. |
52 |
|
53 |
--- |
54 |
[1] I'm running ntp and the initial ntp-client connection and time sync |
55 |
takes ~12 seconds a lot of the time, just over the initial 10 seconds |
56 |
down, 50 to go, trigger on openrc's 1-minute timeout. |
57 |
|
58 |
[2] 131 packages in @live-rebuild, with kde-live-branch, currently |
59 |
4.10.49.9999, being low 120-some, plus choice bits of of X/mesa/drivers |
60 |
and a few other packages including openrc, btrfs-progs and pan. The |
61 |
@live-rebuild typically takes ~20 minutes with ccache, a good portion of |
62 |
which is kdelibs, so the whole update including the sync and change/git- |
63 |
logs check for interesting stuff, @world update, @live-rebuild, etc- |
64 |
update and revdep-rebuild/depclean, runs ~1 hour, often less, sometimes |
65 |
more if there's a new mozilla-overlay firefox build or something in |
66 |
addition to the kde-libs long-build update. |
67 |
|
68 |
[3] Also relevant, 16 gigs RAM, PORTAGETMPDIR on tmpfs. |
69 |
|
70 |
-- |
71 |
Duncan - List replies preferred. No HTML msgs. |
72 |
"Every nonfree program has a lord, a master -- |
73 |
and if you use the program, he is your master." Richard Stallman |