-
Notifications
You must be signed in to change notification settings - Fork 667
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Revert "[sonic-package-manager] support sonic-cli-gen and packages wi… #1972
Conversation
…th YANG model (sonic-net#1650)" This reverts commit f5e5a56.
@stepanblyschak i had to revert your PR1650 to fix this build error https://dev.azure.com/mssonic/be1b070f-be15-4154-aade-b1d3bfb17054/_apis/build/builds/57684/logs/141. Please check your changes against latest master build. |
@prgeor The failure indicates that there is an issue with init_cfg.json, the problem is that init_cfg.json is not aligned to YANG. The failure is not related to an issue in the code change in 1650,
This means that (https://github.com/Azure/sonic-buildimage/blob/master/files/build_templates/init_cfg.json.j2#L39):
is not accepted by YANG validation, which makes sense because according to YANG model for FEATURE table (https://github.com/Azure/sonic-buildimage/blob/master/src/sonic-yang-models/yang-models/sonic-feature.yang#L17) the "state" can only be "enabled", "disabled", "always_enabled", "always_disabled". I am not familiar why the change that made "state" value a jinja template, but if that the new schema for it then we need to fix YANG model for FEATURE table and make the "state" field of type "string". |
@shlomibitton @stepanblyschak Could you discuss and identify which component need fix? |
@qiluo-msft I created this PR sonic-net/sonic-buildimage#9587 that I propose as a fix. To fix this, my PR relaxes the constraint on the "state" field, so it can be any string, as it might be difficult to validate that the "state" field is at least syntactically valid jinja template in the YANG model. On the other hand, the YANG validation will not raise any errors on really incorrect "state" field values - so that's my concern. I am wondering why sonic-net/sonic-buildimage#6700 was the proposed solution and what is the use case? It looks like it complicates the schema a lot and it is kind of introducing a new concept of dynamic configuration in sonic, it also compliates the hostcfgd. Couldn't it be solved on the configuration generation tools (minigraph, sonic-cfggen). It is much simpler and makes more sense - if the config we are generating is for the DualTor then enable mux service if not make it always disabled. Is there an HLD for this feature? Maybe we can discuss about a different solution instead of relaxing YANG validation? Unfortunatelly, the PR that is beeing reverted by this PR was developed half a year ago when there was no YANG/config DB issues but it got merged recently and exposed all those issues introduced since then. |
@stepanblyschak for now i need to roll back your pr since more than a dozen PRs are awaiting to be available in master for more than a month now - sonic-net/sonic-buildimage#9467 |
@qiluo-msft @stepanblyschak |
…kages with YANG model (sonic-net#1650)" (sonic-net#1972)" This reverts commit fe00bbf.
…kages with YANG model (sonic-net#1650)" (sonic-net#1972)" This reverts commit fe00bbf.
…kages with YANG model (#1650)" (#1972)" (#1994) This reverts commit fe00bbf. - What I did Revert previous revert, since the proposed fix has been merged - sonic-net/sonic-buildimage#9587 - How I did it Revert the revert. - How to verify it Run build an on the switch.
…kages with YANG model (#1650)" (#1972)" (#1994) This reverts commit fe00bbf. - What I did Revert previous revert, since the proposed fix has been merged - sonic-net/sonic-buildimage#9587 - How I did it Revert the revert. - How to verify it Run build an on the switch.
…kages with YANG model (sonic-net#1650)" (sonic-net#1972)" (sonic-net#1994) This reverts commit fe00bbf. - What I did Revert previous revert, since the proposed fix has been merged - sonic-net/sonic-buildimage#9587 - How I did it Revert the revert. - How to verify it Run build an on the switch.
Revert #1650
This reverts commit f5e5a56.
Getting following build failure while updating sonic-utilities submodule PR9467