[TSDR-31] Fix broken tsdr feature failing the new extended SingleFeatureTest incl. TestBundleDiag due to ClassNotFoundException: odlparent.bundlestest.TestBundleDiag Created: 18/Jan/17 Updated: 22/Mar/18 Resolved: 22/Mar/18 |
|
| Status: | Resolved |
| Project: | tsdr |
| Component/s: | General |
| Affects Version/s: | unspecified |
| Fix Version/s: | None |
| Type: | Bug | ||
| Reporter: | Michael Vorburger | Assignee: | Scott Melton |
| Resolution: | Cannot Reproduce | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
Operating System: All |
||
| Issue Links: |
|
||||||||||||
| External issue ID: | 7588 | ||||||||||||
| Description |
|
The goal of this issue is to fix whatever caused a "ClassNotFoundException: odlparent.bundlestest.TestBundleDiag" when the new extended SingleFeatureTest incl. TestBundleDiag attempts to install the tsdr feature named "odl-hbaseclient", and remove the exclusion from the BLACKLISTED_BROKEN_FEATURES in SingleFeatureTest in odlparent. Note that TestBundleDiag is a new class in the bundles-test in odlparent introduced as part of https://lists.opendaylight.org/pipermail/dev/2017-January/003106.html. In all other projects, this works (can be found); most likely reason for it not working in tsdr must be that it creates it feature in an old way, and not use the odlparent feature-parent. |
| Comments |
| Comment by Vratko Polak [ 10/Mar/17 ] |
|
Good news and bad news. Testing locally, odl-hbaseclient seems to pass SingleFeatureTest even when removed from the blacklist. The fix [0] is already merged. But, a similar issue has appeared. Not sure if the issue is identical, as I cannot reproduce the original Tsdr exception. The issue is caused by a specific patch set [1] for Yangtools, which causes SFT failure for odl-distribution-version (which is bad as that happens in distribution-check job, affecting every project). See the stack trace below. The latest hypothesis (from Robert) is that the NoClassDefFoundError is a sign of java package name collision, in this case a leftover from Section from surefire log [2]: 2017-03-09 13:50:47,287 | ERROR | ime-limited test | ConfigManagerActivator | 181 - org.opendaylight.controller.config-manager - 0.6.0.SNAPSHOT | Error starting config manager [0] https://git.opendaylight.org/gerrit/53138 |
| Comment by Robert Varga [ 16/Mar/17 ] |
|
Reverted in https://git.opendaylight.org/gerrit/53372 |
| Comment by Michael Vorburger [ 06/Apr/17 ] |
|
https://git.opendaylight.org/gerrit/#/c/53109/ probably fixes the problem this bug originally was about (first comment above) under Karaf 4 as a "side effect", so once that change will finally be merged, it would be worthwhile for someone interestd in TSDR (that's not me...) to retry this, and then probably be able to remove the black listed odl-hbaseclient from tsdr from SingleFeatureTest in features4-test. This is specific to Karaf 4 because of c/53109; under Karaf 3 in features-test it's likely still broken, but maybe we can just forget about that. Vratko / Robert, I don't think I understand what your point above has to do with this; to me that problem (NoClassDefFoundError: org/opendaylight/mdsal/binding/generator/util/BindingRuntimeContext) looks like a completely separate issue. |
| Comment by Robert Varga [ 07/Apr/17 ] |
|
The BindingRuntimeContext thing was a split package problem, fixed in https://git.opendaylight.org/gerrit/53140. |
| Comment by Tom Pantelis [ 22/Mar/18 ] |
|
This is no longer an issue - odl-hbaseclient SFT runs fine now. |