1 |
Hi Arve, |
2 |
|
3 |
On 2022/08/30 12:27, Arve Barsnes wrote: |
4 |
|
5 |
> On Tue, 30 Aug 2022 at 11:52, Jaco Kroon <jaco@××××××.za> wrote: |
6 |
>> I note that the default provider for the virtual is systemd-utils[udev], |
7 |
>> followed by sys-fs/udev, sys-fs/eudev and finally sys-apps/systemd. |
8 |
>> This contradicts the wiki page which states that sys-fs/udev is the |
9 |
>> default (https://wiki.gentoo.org/wiki/Udev). |
10 |
>> |
11 |
>> Is there more comprehensive documentation around about this, and which |
12 |
>> should be preferred when? |
13 |
> sys-fs/udev is also just a virtual now, which pulls in |
14 |
> systemd-utils[udev], so using either works exactly the same. |
15 |
|
16 |
Thanks, I missed that. That does help to clear things up. |
17 |
|
18 |
This leaves three implementations, systemd-utils[udev], eudev and systemd. |
19 |
|
20 |
We don't use systemd so that eliminates that, can I safely assume that |
21 |
systemd-utils[udev] is "extracted" from systemd and really the same |
22 |
thing? Ie, it's the udevd without the associated systemd environment? |
23 |
So really only two implementations. |
24 |
|
25 |
eudev then is the wild horse in that it was forked from the days when |
26 |
sys-fs/udev got incorporated into systemd, and have been following a |
27 |
parallel but somewhat independent path? It doesn't contain many of the |
28 |
newer systemd-udev "features" and "lags behind"? |
29 |
|
30 |
Which, assuming then my understanding is now improved (which I believe |
31 |
it is), then the selection should be based as follows: |
32 |
|
33 |
1. If you're using systemd, use the embedded udevd. |
34 |
2. If you're using openrc, you should prefer sys-fs/udev aka |
35 |
systemd-utils[udev] unless you have a specific reason to rather use eudev? |
36 |
|
37 |
eudev has served me very well for very long, and avoided a fair number |
38 |
of LVM related deadlock issues we experienced with sys-fs/udev at the |
39 |
time. We've been moving back, and I'm not convinced those have been |
40 |
eliminated, but I could not yet prove any of the recent system deadlocks |
41 |
we've seen relates to systemd-udev. |
42 |
|
43 |
(The one deadlock we did manage to trap was without a doubt something |
44 |
with the linux kernel IO scheduler, possibly related to raid6, however, |
45 |
lvm is also involved in that path, which does involve udev. Also the |
46 |
system that most frequently runs into problems - only one with software |
47 |
raid6, but it also by far makes the most aggressive use of lvm |
48 |
snapshots. Thus no definitive patterns.) |
49 |
|
50 |
Being a creature of habit based on experience I am sceptical. I am |
51 |
contemplating throwing that one host back to eudev and seeing if that |
52 |
"solves" the problem ... but how long is a piece of string. |
53 |
|
54 |
Thanks for the information, I'll let the above simmer a good long while. |
55 |
|
56 |
Kind Regards, |
57 |
Jaco |