[NETCONF-113] Incorrect Yang models should not prevent functionality from unrelated components Created: 21/Dec/15  Updated: 04/May/20

Status: Confirmed
Project: netconf
Component/s: restconf-nb
Affects Version/s: None
Fix Version/s: None

Type: Bug
Reporter: Vratko Polak Assignee: Unassigned
Resolution: Unresolved Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Operating System: All
Platform: All

External issue ID: 4828

 Description   

In Beryllium, all jobs were failing due to Restconf getting stuck in http code 401.
Investigation suggests it is not an AAA bug, it is Yangtools unable to move on when encountering invalid Yang modules.

The situation with all jobs can be solved in integration/distribution by removing offending features; but as ODL grows, it would be good to make the core more resistant to low-quality components.

Search for "Failed to parse YANG from source" in (long) [0] to see current offending projects, see [1] for the integration/distribution fix (which perhaps will be merged as 31720 instead).

[0] https://jenkins.opendaylight.org/releng/view/bgpcep/job/bgpcep-csit-1node-periodic-throughpcep-all-beryllium/83/consoleFull
[1] https://git.opendaylight.org/gerrit/31715

 Comments   
Comment by Robert Varga [ 08/Jan/16 ]

All interfaces required for RESTCONF to exclude specified models are available, making this a non-yangtools issue.

Note that, though, that should this really be implemented, RESTCONF has to emit a very stern warning and (some) test jobs have to check for this warning and fail tests. Otherwise this problem will not be detected and we will end up shipping a broken release.

Comment by Vratko Polak [ 08/Jan/16 ]

> (some) test jobs have to check for this warning

Agreed. We can add such check to deploy test job (which is planned to be run during

{project}

merge* job execution).
That should cover everything under odl-integration-all.

See these two lines [0] for what is checked already (the code may be move someplace else later).

[0] https://git.opendaylight.org/gerrit/gitweb?p=releng/builder.git;a=blob;f=jjb/integration/include-raw-integration-deploy-controller-verify.sh;h=ff1fd97a60fdf8c6eac7c927ca93c14ed8d1b103;hb=HEAD#l97

Comment by Jamo Luhrsen [ 04/May/20 ]

rovarga, this is a really old ticket at this point. Are we past this kind of problem now, or
something more to do here. If I understand correctly, the issue is that a bad yang model
would bork feature installation and break the entire distribution. Originally our CSIT was
noticing the symptom of 401 unauthorized. Are we still at risk of this?

Generated at Wed Feb 07 20:14:12 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.