[CONTROLLER-1877] odl-mdsal-trace feature prevents ODL from starting in stable/fluorine Created: 12/Dec/18  Updated: 18/Dec/18  Resolved: 18/Dec/18

Status: Resolved
Project: controller
Component/s: None
Affects Version/s: Neon, Fluorine SR1
Fix Version/s: Neon, Fluorine SR2

Type: Bug Priority: Medium
Reporter: Victor Pickard Assignee: Michael Vorburger
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Attachments: File karaf.log     File karaf.log    

 Description   

In stable/fluorine, loading the following features prevents ODL from starting. Lots of logs where bundles are failing to initialize, waiting on dependencies.

odl-netvirt-openstack
odl-mdsal-trace

If we just load odl-netvirt-openstack, ODL will start fine.

 Comments   
Comment by Robert Varga [ 12/Dec/18 ]
  1. What are the steps to reproduce?
  2. Can we get those logs?
  3. Is the issue reproducible on current neon?
Comment by Robert Varga [ 12/Dec/18 ]

odl-mdsal-trace is a controller feature.

Comment by Victor Pickard [ 12/Dec/18 ]

The steps to reproduce are to load odl-mdsal-trace and odl-netvirt-openstack at the same time in stable/fluorine branch.

I added the logs.

I don't know if this is an issue in neon. Encountered this issue while doing some downstream testing for stable/oxygen and stable/fluorine.

Comment by Victor Pickard [ 12/Dec/18 ]

Thanks.

Comment by Michael Vorburger [ 13/Dec/18 ]

The attached log shows daexim and genius failing in BP due to this:

Caused by: java.lang.NullPointerException
	at org.opendaylight.mdsal.binding.dom.codec.impl.BindingCodecContext.getCodecContextNode(BindingCodecContext.java:151) ~[?:?]
	at org.opendaylight.mdsal.binding.dom.codec.impl.InstanceIdentifierCodec.deserialize(InstanceIdentifierCodec.java:36) ~[?:?]

Not sure how the installation of odl-mdsal-trace causes this, needs local repro and debugging...

There are other exceptions in this log as well, but those are probably just impacts; I suggest we ignore and focus on above.

There is also the remote infratuils.diagstatus JMX failing, that may be an unrelated problem.

Comment by Michael Vorburger [ 14/Dec/18 ]

So this can easily be locally reproduced, I just built today's master (Neon, which is also affected, not just stable Fluorine; notthing much really change in this code since Oxygen BTW, so this is curious) netvirt/karaf, and do this and then immediately hit it, during boot, without having to even "run" anything:

opendaylight-user@root>feature:install odl-mdsal-trace 
opendaylight-user@root>feature:install odl-netvirt-openstack

Attached is the full karaf.log from such a local run.

Comment by Robert Varga [ 14/Dec/18 ]

Does installing both at in one go manifest the problem, too?

Comment by Michael Vorburger [ 14/Dec/18 ]

Wait a second... the problem is actually very obvious - I think. (Not sure how it could have broken though).

I'm locally testing a fix and if that goes well should have a Gerrit coming up.

Comment by Michael Vorburger [ 14/Dec/18 ]

> notthing much really change in this code since Oxygen BTW, so this is curious
> Not sure how it could have broken though

https://git.opendaylight.org/gerrit/#/c/72658 is probably the root cause that lead to this issue.

Generated at Wed Feb 07 19:56:40 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.