GNU bug report logs - #76081
OCI provisioning service

Previous Next

Full log

View this message in rfc822 format

From: paul <goodoldpaul <at> autistici.org>
To: 76081 <at> debbugs.gnu.org
Subject: [bug#76081] OCI provisioning service
Date: Mon, 5 May 2025 09:54:49 +0200

Hi,

On 3/18/25 18:50, paul wrote:
> Hi,
>
> On 3/9/25 01:27, paul wrote:
>> Hi,
>>
>> On 3/4/25 13:39, paul wrote:
>>> Hi,
>>>
>>> On 2/27/25 00:01, paul wrote:
>>>> Hi guix,
>>>>
>>>> On 2/18/25 02:20, paul wrote:
>>>>> Hi,
>>>>>
>>>>> On 2/12/25 02:09, paul wrote:
>>>>>> Hi guix,
>>>>>>
>>>>>> On 2/9/25 21:38, paul wrote:
>>>>>>> I'm sending a v3 fixing a bug in the merge algorithm for volumes 
>>>>>>> and networks.
>>>>>>>
>>>>>>> On 2/9/25 20:14, paul wrote:
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> I'm about to send a v2. v2 compared to the first revision 
>>>>>>>> features:
>>>>>>>>
>>>>>>>>
>>>>>>>> - it actually compiles all the times :) (rev 1 referenced 
>>>>>>>> oci-image too early for it to be working and generated a 
>>>>>>>> compile time error, if you recompiled it sometimes went away so 
>>>>>>>> I thought it was a problem of my setup. CI caught this)
>>>>>>>> - it allows more values to be overridden by eventual users of 
>>>>>>>> the Scheme API
>>>>>>>> - it allows passing extra arguments directly after each podman 
>>>>>>>> or docker invokation, allowing for example for overriding 
>>>>>>>> podman --root and similar options.
>>>>>>>>
>>>>>>>> All of these tests should pass:
>>>>>>>>
>>>>>>>> guix shell -D guix -CPW -- make check-system 
>>>>>>>> TESTS="oci-container oci-service-rootless-podman docker 
>>>>>>>> docker-system rootless-podman oci-service-docker"
>>>>>>>>
>>>>>> I'm sending a v4 changing slightly the image loader, the same 
>>>>>> tests as before are supposed to pass. Now the Home service [0] is 
>>>>>> working for me with rootless podman. I'll try it on different 
>>>>>> distros if I manage to.
>>>>>
>>>>> I'm sending a v5 implementing a Home service. The changes compared 
>>>>> to v4 are pretty trivial as the plumbing was already there, the 
>>>>> only downside is that I'm not able to use for-home? in 
>>>>> define-configuration, so I had to reimplement oci-configuration 
>>>>> with (guix records) and had to reimplement some validation (gnu 
>>>>> services configuration) would figure out magically.
>>>>
>>>> I'm sending  a v6. Compared to v5 it resolves the conflicts with 
>>>> master and it should fix the stop action for rootless podman backed 
>>>> services.
>>>
>>> I'm sending a v7 fixing the restart action for podman backed services.
>>
>> I'm sending a v8 fixing some further bugs wrt to the start and stop 
>> of podman backed services.
>
> In the hope of making the review easier, I'm sending a v9 that is 
> completely the same of v8 in terms of changes but with smaller 
> commits. Hopefully this makes it easier to review changes, please let 
> me know if I can do anything to ease the workload for reviewers of 
> this patch set.

I'm sending a v10 rebased on current master. The only difference with v9 
is that this revision exports the 
oci-{container,network,volume}-configuration->options procedures , that 
can be useful when writing services that extend the oci-service-type.


cheers,

giacomo

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.