1 |
W dniu wto, 12.05.2020 o godzinie 10∶05 -0700, użytkownik Zac Medico |
2 |
napisał: |
3 |
> On 5/12/20 3:39 AM, Michał Górny wrote: |
4 |
> > I'm sorry but I was asking of a more high-level implications. |
5 |
> > |
6 |
> > I presume that this means that more than files of more than one |
7 |
> > package |
8 |
> > can be merged simultaneously. However: |
9 |
> |
10 |
> No, an exclusive lock must be held on vardbapi._fs_lock for this. |
11 |
> This |
12 |
> is currently required at least to guarantee that access to the config |
13 |
> memory file is serialized (config memory is the thing that emerge |
14 |
> --noconfmem disables, but --noconfmem does not disable this lock). |
15 |
> |
16 |
> We assume that it's probably not worthwhile to try to merge files for |
17 |
> more than one package at a time, since that would cause them to |
18 |
> compete |
19 |
> for IO bandwidth. |
20 |
> |
21 |
> > 1. Are collisions handled correctly then? i.e. if you start |
22 |
> > installing |
23 |
> > A, and then B, and the two packages collide will portage fail |
24 |
> > before |
25 |
> > starting to install any file from B? |
26 |
> |
27 |
> There are no guarantees here. However, the risk is minimal, since |
28 |
> it's |
29 |
> unlikely that a file collision of this sort would occur. file |
30 |
> collisions |
31 |
> are a QA problem that is generally detected and corrected log before |
32 |
> we |
33 |
> would encounter a collision of this sort. |
34 |
> |
35 |
> > 2. Are preinst/postinst phases called simultaneously or serialized? |
36 |
> |
37 |
> They're serialized. |
38 |
|
39 |
Now I'm lost here. Could you try to explain to me, without getting |
40 |
into the deep technicalities, how parallel-install achieves better |
41 |
speed or at doing what is non-parallel-install so slow? |
42 |
|
43 |
-- |
44 |
Best regards, |
45 |
Michał Górny |