🐛🚑️ Make director-v0 work with registry:3.0.0
#8239
Conversation
mrnicegyu11
requested review from
GitHK,
pcrespov and
sanderegg
as code owners
mrnicegyu11
added
bug
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #8239 +/- ##
==========================================
- Coverage 88.02% 88.01% -0.01%
==========================================
Files 1916 1916
Lines 74053 74053
Branches 1301 1301
==========================================
- Hits 65184 65180 -4
- Misses 8478 8482 +4
Partials 391 391
Continue to review full report in Codecov by Sentry.
🚀 New features to boost your workflow:
|
|
| "application/vnd.oci.image.manifest.v1+json", | ||
| "application/vnd.oci.image.index.v1+json", |
There was a problem hiding this comment.
Not fully convinced about this approach. What about addressing the issue in the code?
I think it's better to use the default version 2 spec instead of requesting the older one I would expect this one to get deprecated at some point.
log_level=ERROR | log_timestamp=2025-08-20 09:01:21,941 | log_source=simcore_service_director.registry_proxy:log_catch(575) | log_uid=None | log_oec=None | log_trace_id=0 | log_span_id=0 | log_msg=Unhandled exception:
Traceback (most recent call last):
File "/home/scu/.venv/lib/python3.11/site-packages/servicelib/logging_utils.py", line 570, in log_catch
yield
File "/home/scu/.venv/lib/python3.11/site-packages/simcore_service_director/registry_proxy.py", line 449, in get_repo_details
if image_details := await image_details_future:
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/scu/.venv/lib/python3.11/site-packages/simcore_service_director/registry_proxy.py", line 402, in get_image_details
labels, image_manifest_digest = await get_image_labels(
^^^^^^^^^^^^^^^^^^^^^^^
File "/home/scu/.venv/lib/python3.11/site-packages/simcore_service_director/registry_proxy.py", line 384, in get_image_labels
v1_compatibility_key = json_loads(request_result["history"][0]["v1Compatibility"])
~~~~~~~~~~~~~~^^^^^^^^^^^
KeyError: 'history'
There was a problem hiding this comment.
The issue is addressed in the code ;) These are code-changes after all. In case you want to do it differently, feel free to take it over in case ;) I interpret your comment as that I should not quickly get this in (I would have done it this way) but wait for some others. Must I wait for Sylvain, or even Pedro?
🧪 CI InsightsHere's what we observed from your CI run for 380e5d3. ❌ Failed Jobs
✅ Passed Jobs With Interesting Signals
|
|
@mrnicegyu11 just created #8240 to ensure the dv-0 is compatible with both types. This was adjusted to the newer schema version. Since multi-arch images might become a thing soonish, I guess the dv0 shall become compatible at some point as well. This would be a first step |
|
Closing in favor of #8240 thanks a bunch @sanderegg , appreciated! |
5 participants
What do these changes do?
Turns out that in some internal calls the dv0 relies on the docker V1 registry protocol. The
registry:3.0.0seems to by default return v2 specs if the client to the docker registry does not specify what is acceptable. Via the usualAcceptheaders we can inform the docker registry even on version 3.0.0 that we would like to recieve only oci-image v1 data, which is compatible with the dv0 code.Bonus:
Update registry to
3.0.0everywhereRelated issue/s
How to test
Tested locally (
make up-devel)Dev-ops