[NETCONF-97] netconf won't come up with 'odl-integration-compatible-with-all' Created: 05/Nov/15 Updated: 15/Mar/19 Resolved: 04/Jan/16 |
|
| Status: | Resolved |
| Project: | netconf |
| Component/s: | netconf |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | ||
| Reporter: | Jozef Behran | Assignee: | Unassigned |
| Resolution: | Done | 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: | 4583 | ||||||||
| Description |
|
Put 'config,standard,region,package,kar,ssh,management,odl-integration-compatible-with-all,odl-restconf,odl-netconf-mdsal,odl-netconf-connector-all' into bootFeatures, then fire ODL up. Netconf will fail to come up. When debug logging was enabled around the time Netconf was expected to be up, the following fragment was spotted in the logs repeating over and over again: ... (the first "Trying to create the platform default provider" at the start is from the previous iteration of the infinite loop, the last one belongs to the current iteration and is followed by another instance of the second log message which is followed by another instance of the third log message and so on and so forth). From my perspective this is what the logs are telling me about what is happening: 1. "/usr/lib/jvm/java-1.7.0-openjdk-1.7.0.75.x86_64/jre/lib/rt.jar!/com/sun/xml/internal/bind/v2/ContextFactory.class" tries to access "com.sun.xml.internal.bind.XmlAccessorFactoryis" |
| Comments |
| Comment by Jozef Behran [ 05/Nov/15 ] |
|
Occurs in stable-lithium branch: |
| Comment by Luis Gomez [ 05/Nov/15 ] |
|
Right we talk about this issue every Li SR reviews, thanks for opening the bug. |
| Comment by Vratko Polak [ 09/Nov/15 ] |
|
Looking at karaf.log [0], there is another symptom, visible due to workaround for 2015-11-05 19:19:46,735 | WARN | r - ConfigPusher | FeatureConfigPusher | 144 - config-persister-feature-adapter - 0.3.3.SNAPSHOT | Feature: standard-condition-webconsole_0_0_0, 3.0.3 is missing from features service. Skipping Perhaps ConfigPushingRunnable does not detect there is nothing more to push; or (more probably) there is still something to push, but there is not enough logging to see what it is and why it does not get pushed quickly enough. |
| Comment by Vratko Polak [ 12/Nov/15 ] |
|
Manual testing shows that if all odl-sfc* features are removed from odl-integration-compatible-with-all, this Bug does not happen. |
| Comment by Vratko Polak [ 12/Nov/15 ] |
|
It is the But there were no changes in sfc or main netconf recently, the change which introduced failures was in fact https://git.opendaylight.org/gerrit/28985 So this seem to be a bug in config persister implementation. |
| Comment by Vratko Polak [ 12/Nov/15 ] |
|
> the change which introduced failures was in fact https://git.opendaylight.org/gerrit/28985 Not true. I looked at wrong job, sorry. |
| Comment by Vratko Polak [ 12/Nov/15 ] |
|
> etc/opendaylight/karaf/sfc-netconf-initial.xml And also this is not a real signal, just a coincidence due to my uneven patience. If this really turns out to be merely a bootup performance bug, perhaps this e-mail would be relevant for the fix: https://lists.opendaylight.org/pipermail/controller-dev/2015-September/010305.html |
| Comment by Vratko Polak [ 12/Nov/15 ] |
|
Turns out my first binary search was focused on the wrong symptom. CPU stability always eventually happen, the critical symptom is whether there are data mounted on controller-config afterwards. |
| Comment by Vratko Polak [ 13/Nov/15 ] |
|
My examination found a single culprit: Workaround to unblock testing without actually fixing |
| Comment by Christine Hsieh [ 04/Jan/16 ] |
|
|