Tuesday triage (2023-02-06)

[Bryce Harrington]

On Mon, Feb 13, 2023 at 06:37:10PM -0300, Athos Ribeiro wrote:
> On Mon, Feb 13, 2023 at 06:17:16PM -0300, Lucas Kanashiro wrote:
> > # Bug triage
> >
> > ## https://pad.lv/1581864 | *      | DXBCDE  | Won't Fix     | nginx
> >        | nginx.service: Failed to read PID from file /run/nginx.pid:… |
> > ## https://pad.lv/2006054 | *      | DBF     | Triaged       | bind9
> >        | Bind9 service file for 20.04 and older is missing Restart p… |
> >
> > Both bugs above have the server-triage-discuss tag (added last Monday), not
> > sure if we already discussed them and if the tag could be removed. Athos
> > may refresh my mind here.
>
> This is from my triage report back then:
>
> https://pad.lv/2006054 | bind9 | Bind9 service file for 20.04 and older is missing Restart p…
>   This is a request to add a "Restart=on-failure" entry in the named systemd
>   unit file. I linked the Debian bug that fixed it in Debian back in 2020. The
>   issue only affects focal and bionic, so this is an SRU request. Since I am
>   not sure if backporting the service change was discussed in the past, I added
>   a server-triage-discuss tag to this one.
>
> Does anyone recall ever discussing this? If there are no constraints,
> this could just be tagged server-todo.
>
> https://pad.lv/1581864 | nginx | nginx.service: Failed to read PID from file /run/nginx.pid:…
>   This is a low priority bug in nginx which only affects bionic. As we approach
>   the end of standard support for that series, I wonder if we should bump the
>   priority to ensure it is dealt with in time or not. I added a
>   server-triage-discuss tag to the bug.
>
> This one is a bit more sensitive IMO. It is about whether we want to
> bump the priority for the bug __because it is in bionic__. If we don't
> do it soon, it will not get fixed in time (EOSS coming). I believe we
> should keep the server-triage-discuss tag here.

Taking a quick look, the patch in question is not from upstream (indeed
upstream appears to have rejected it on basis it looks wrong to them).

The impact of the issue appears to be that an error message is logged,
however if I understand correctly it does not cause any behavioral
changes to the program.  If this is true, the [Impact] here seems pretty
low.

A workaround is identified in the bug description, that looks pretty
straightforward.  Alternative workarounds are discussed in the comments.

There had been some healthy discussion on this while bionic was the
current LTS, but engagement from users and reporters seems to have ended
after focal's release.  Only comments since then has been ourselves and
Simon Déziel (the original reporter).


So, I'm thinking maybe no one is really in need of this fix at this
point as they've likely either upgraded or applied the workaround or
don't care about the innocuous error message.  Plus, the patch sounds
like it needs to be redone, and that would bring some non-zero amount of
risk.

Other thoughts?

Bryce