[ODLPARENT-77] [Karaf 4] SFT ChannelException: Failed to bind to: /127.0.0.1:2550 Created: 08/Mar/17 Updated: 24/Jan/18 Resolved: 20/Apr/17 |
|
| Status: | Resolved |
| Project: | odlparent |
| Component/s: | General |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | ||
| Reporter: | Michael Vorburger | Assignee: | Unassigned |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
Operating System: All |
||
| Attachments: |
|
||||||||
| Issue Links: |
|
||||||||
| External issue ID: | 7926 | ||||||||
| Description |
|
when running SingleFeatureTest (SFT) on a Karaf 4 feature, e.g. by "mvn -o -Dsft.diag.skip=true clean install" in genius/features/features4-genius from https://git.opendaylight.org/gerrit/#/c/51966/ (not sure if this also happens in production Karaf or is SFT specific), see full SFT log attached to (otherwise probably unrelated) then I'm seeing this, even though I'm pretty sure I have nothing running on port 2550, so it looks like we perhaps somewhere don't shut down a Netty from a previous feature install - kind of thing? Do we need to do something like I've done (unrelated) in https://git.opendaylight.org/gerrit/#/c/52976/ in .. where- where is this ForkJoinPool (below) which launches a NettyTransport started? 2017-03-08 00:14:02,912 | ERROR | pool-2-thread-1 | ServiceRecipe | 14 - org.apache.aries.blueprint.core - 1.6.2 | Error retrieving service from ServiceRecipe[name='.component-1'] |
| Comments |
| Comment by Robert Varga [ 13/Mar/17 ] |
|
This is most probably caused by SFT running concurrently because of rewiring happening. Please check bundle startup vs. test sequencing |
| Comment by Michael Vorburger [ 29/Mar/17 ] |
|
Just ran into this again in https://jenkins.opendaylight.org/releng/job/genius-verify-carbon-mvn33-openjdk8/1997/console for https://git.opendaylight.org/gerrit/#/c/51966/9 ... see attached bug7926.txt. |
| Comment by Michael Vorburger [ 29/Mar/17 ] |
|
Attachment bug7926.txt has been added with description: Stack trace of this error from genius Karaf 4 SFT build |
| Comment by Michael Vorburger [ 29/Mar/17 ] |
|
NB: NETIDE-7 seems to say this may be related to particular features, reproducible even outside of SFT, if (quote) 'to reproduce this first install "odl-vpnservice-openstack" then install "odl-netide-rest"' - exception seen on Karaf console, not log file. |
| Comment by Robert Varga [ 29/Mar/17 ] |
|
This is a classic problem: the test container is being rewired: 2017-03-29 10:29:38,409 | INFO | ime-limited test | FeaturesServiceImpl | 8 - org.apache.karaf.features.core - 4.0.7 | Adding features: odl-genius-api/[0.2.0.SNAPSHOT,0.2.0.SNAPSHOT] 2017-03-29 10:29:40,568 | INFO | pool-2-thread-1 | FeaturesServiceImpl | 8 - org.apache.karaf.features.core - 4.0.7 | Bundles to install: 2017-03-29 10:29:41,214 | INFO | pool-2-thread-1 | FeaturesServiceImpl | 8 - org.apache.karaf.features.core - 4.0.7 | Stopping bundles: |
| Comment by Robert Varga [ 29/Mar/17 ] |
|
Also the generated features do not look right: shows odl-genius-fcaps-framework expanding things that it should be taking from controller/mdsal features. |
| Comment by Michael Vorburger [ 06/Apr/17 ] |
|
I'm not seeing the ChannelException error shown above anymore in target/surefire-reports/org.opendaylight.odlparent.featuretest.SingleFeatureTest-output.txt when running "mvn clean package" in genius/features/features4-genius under the latest patch set 15 of c/53109 ... not sure why, I didn't actually do anything in genius since this - has someone else silently fixed it? Robert, I'm not clear what the point you raised above means, and how it relates to the error shown in this bug... do you want to propose a Gerrit to change the feature definition? Otherwise, do you have any objection to closing this bug now? |
| Comment by Michael Vorburger [ 20/Apr/17 ] |
|
> I'm not seeing the ChannelException error shown above anymore in this still happens sporadically, e.g. again on https://logs.opendaylight.org/releng/jenkins092/genius-verify-carbon-mvn33-openjdk8/1997/archives/features/features4-genius/target/feature/feature.xml.gz > shows odl-genius-fcaps-framework expanding things that Stephen has said on IRC he'll try to push a change to address this.. |
| Comment by Robert Varga [ 20/Apr/17 ] |
|
'expanding' means it is pulling in and installing bundles explicitly instead of relying on upstream feature to provide them. I think the magic dance is to add scope=runtime dependency on appropriate feature, but I always forget what it actually is. |
| Comment by Stephen Kitt [ 20/Apr/17 ] |
|
Current builds are correct, using odl-mdsal-broker:
|
| Comment by Michael Vorburger [ 20/Apr/17 ] |
|
This actually happens quite a lot, just noticed that e.g. https://jenkins.opendaylight.org/releng/job/genius-verify-nitrogen-mvn33-openjdk8/95/console for https://git.opendaylight.org/gerrit/#/c/54374/ also failed because of this... |
| Comment by Michael Vorburger [ 20/Apr/17 ] |
|
as per IRC discussion on #opendaylight today: https://git.opendaylight.org/gerrit/#/c/55758/ should fix the root cause. https://git.opendaylight.org/gerrit/#/c/55759/ should prevent similar problems. Closing as RESOLVED / FIXED - not verified; can/will re-open if still seen. |