| 1 |
On 5/12/20 3:39 AM, Michał Górny wrote: |
| 2 |
> W dniu wto, 12.05.2020 o godzinie 01∶40 -0700, użytkownik Zac Medico |
| 3 |
> napisał: |
| 4 |
>> On 5/11/20 10:54 PM, Michał Górny wrote: |
| 5 |
>>> W dniu nie, 10.05.2020 o godzinie 21∶32 -0700, użytkownik Zac |
| 6 |
>>> Medico |
| 7 |
>>> napisał: |
| 8 |
>>>> The feature enables finer grained locks for install operations, |
| 9 |
>>>> and |
| 10 |
>>>> everyone agrees that it's safe to enable by default. |
| 11 |
>>> |
| 12 |
>>> Who's 'everyone' and where's their analysis of the problem? |
| 13 |
>>> The manpage doesn't really help understand what this does, exactly. |
| 14 |
>> |
| 15 |
>> Before parallel-install there was just one big lock, so only one |
| 16 |
>> package |
| 17 |
>> slot could enter the merge/unmerge state at a given time. |
| 18 |
>> |
| 19 |
>> With parallel install, there are a few finer-grained locks that come |
| 20 |
>> into play. [...] |
| 21 |
> |
| 22 |
> I'm sorry but I was asking of a more high-level implications. |
| 23 |
> |
| 24 |
> I presume that this means that more than files of more than one package |
| 25 |
> can be merged simultaneously. However: |
| 26 |
|
| 27 |
No, an exclusive lock must be held on vardbapi._fs_lock for this. This |
| 28 |
is currently required at least to guarantee that access to the config |
| 29 |
memory file is serialized (config memory is the thing that emerge |
| 30 |
--noconfmem disables, but --noconfmem does not disable this lock). |
| 31 |
|
| 32 |
We assume that it's probably not worthwhile to try to merge files for |
| 33 |
more than one package at a time, since that would cause them to compete |
| 34 |
for IO bandwidth. |
| 35 |
|
| 36 |
> 1. Are collisions handled correctly then? i.e. if you start installing |
| 37 |
> A, and then B, and the two packages collide will portage fail before |
| 38 |
> starting to install any file from B? |
| 39 |
|
| 40 |
There are no guarantees here. However, the risk is minimal, since it's |
| 41 |
unlikely that a file collision of this sort would occur. file collisions |
| 42 |
are a QA problem that is generally detected and corrected log before we |
| 43 |
would encounter a collision of this sort. |
| 44 |
|
| 45 |
> 2. Are preinst/postinst phases called simultaneously or serialized? |
| 46 |
|
| 47 |
They're serialized. |
| 48 |
-- |
| 49 |
Thanks, |
| 50 |
Zac |
Archives