[bug#30637] [WIP] shepherd: Poll every 0.5s to find dead forked services

[Carlo Zancanaro <carlo@zancanaro.id.au>]

[-- Attachment #1: Type: text/plain, Size: 1962 bytes --]

Hey Ludo,

On Fri, Mar 02 2018, Ludovic Courtès wrote:
>> I am doing that. The problem is that when a service dies 
>> (crashes, quits, etc.) the `respawn?` option cannot be honoured 
>> because shepherd is not notified that the process has 
>> terminated (because it never receives a SIGCHLD for the forked 
>> pid). My patch polls for the processes we expect, to make up 
>> for the lack of notification.
>
> I see.
>
> Actually, thinking more about it, we should be using 
> PR_SET_CHILD_SUBREAPER from prctl(2), which is designed exactly 
> for that.

Excellent! This is exactly the information that I needed. This is 
what I've been looking for, but without enough knowledge to be 
able to find it. Thanks!

> So what about this plan:
>
>   1. Add FFI bindings in (shepherd system) for prctl(2). We 
>   should arrange for it to throw to 'system-error when the 
>   ‘prctl’ symbol is missing, as is the case on GNU/Hurd.

Are we okay with having this just not work on GNU/Hurd (or kernels 
older than 3.4, according to the prctl manpage)? We could fall 
back to a polling approach if prctl isn't available? I don't 
really like the idea of this working on some kernels but not 
others, given that process supervision is one of the main jobs of 
shepherd.

>   2. Use prctl/PR_SET_CHILD_SUBREAPER in ‘exec-command’. Here we 
>   must ‘catch-system-error’ around that call to cater to 
>   GNU/Hurd.

Why would we need to set it in exec-command? It looks like it 
modifies the state of the calling process, which means we'd want 
to set it in the shepherd service, not in each of the child 
processes.

> That would address the main issue without having to resort to 
> polling. Respawning will work only when #:pid-file is used 
> though, but that’s already an improvement.
>
> Thoughts?

I'll try to get this working in the next few days. Hopefully 
you'll see a patch from me soon.

Carlo

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]