[ODLPARENT-110] [Karaf 4] Avoid bundle refreshes Created: 02/Aug/17 Updated: 02/May/18 Resolved: 02/May/18 |
|
| Status: | Resolved |
| Project: | odlparent |
| Component/s: | General |
| Affects Version/s: | 2.0.5 |
| Fix Version/s: | 2.0.6 |
| Type: | Bug | ||
| Reporter: | Vratko Polak | Assignee: | Robert Varga |
| 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: | 8923 | ||||||||||||
| Description |
|
This is an umbrella item, feel free to open child items when it becomes clear which refreshes should be avoided and how. Opening against Odlparent, as that project provided fix for The first issue is with adding Groupbasedpolicy features to Distribution. Unfortunately, I was not able to repeat the failure reliably. First, SingleFeatureTest frequently runs out of heap space, and the exact symptoms vary between runs. I have used this [0] Odlparent patch to increase heap space for SFT JVM, and here is a segment related to refreshes as seen in surefire karaf log [1]. 2017-08-02 11:20:01,180 | INFO | pool-2-thread-1 | FeaturesServiceImpl | 8 - org.apache.karaf.features.core - 4.0.9 | org.apache.karaf.shell.ssh/4.0.9 Note that one refresh is still related to bcpkix even though that should have been avoided by [2]. By the way, how can I tell which commits on Odlparent 2.0.x branch are included in latest (2.0.0) release? [0] https://git.opendaylight.org/gerrit/#/c/61034/1 |
| Comments |
| Comment by Vratko Polak [ 03/Aug/17 ] |
|
Another example, this time from Bier adding features to Distribution [3]: Refreshing bundles: |
| Comment by Robert Varga [ 03/Aug/17 ] |
|
There are a few problems here:
also ssh being refreshed, but I am not sure why that is. The first two problems need a fix in odlparent. |
| Comment by Robert Varga [ 04/Aug/17 ] |
|
The ssh refresh (and bc instsall) should be fixed by this: https://git.opendaylight.org/gerrit/61175 Netty reload will need and odlparent release and have it integrated into all projects using netty. Required odlparent patch is: https://git.opendaylight.org/gerrit/61174. After odlparent is released, https://git.opendaylight.org/gerrit/61177 adjusts mdsal to not package javassist itself. |
| Comment by Robert Varga [ 04/Aug/17 ] |
|
The ssh reload should be fixed. The netty reload should be fixed once odlparent-2.0.3 is released and adopted. I am not sure it makes sense to also address pax-logging and jasypt – I think they are mostly harmless. |
| Comment by Vratko Polak [ 07/Aug/17 ] |
|
This is the patch which needs this Bug fixed to pass verify: [7]. > The netty reload should be fixed once odlparent-2.0.3 is released and adopted. I believe that is true. Sandbox run [8] confirms that, but it also included heap increase for SFT [9]. [7] https://git.opendaylight.org/gerrit/59755 |
| Comment by Robert Varga [ 07/Aug/17 ] |
|
Still needs more patches. |
| Comment by Robert Varga [ 07/Aug/17 ] |
| Comment by Vratko Polak [ 07/Aug/17 ] |
|
> Sandbox run [8] confirms that I take that back, the job was (mis)configured to skip SFT. [11] https://jenkins.opendaylight.org/sandbox/job/integration-multipatch-test-nitrogen/11/ |
| Comment by Vratko Polak [ 07/Aug/17 ] |
|
> Run #11 [11] shows bunch od bundles refreshing The only good news is that the current behavior does not depend on amount of heap space (1 GB or 4 GB) for SFT for the Groupbasedpolicy features. |
| Comment by Robert Varga [ 07/Aug/17 ] |
|
odlparent patches: |
| Comment by Robert Varga [ 07/Aug/17 ] |
|
Manual testing of the combination of netconf patch, the two odlparent patches and a rebuilt aaa seems to look okay. The only refresh I am seeing now is: efreshing bundles: I think we should deal with these at some point, but certainly not before 2.0.3. |
| Comment by Robert Varga [ 07/Aug/17 ] |
|
I forgot to note: mdsal was rebuilt, and the refresh report is from test output of all SFTs in netconf. |
| Comment by Vratko Polak [ 16/Aug/17 ] |
|
I have encountered a new refresh while testing Bier features on Sandbox [14]. 2017-08-16 10:49:43,982 | INFO | pool-2-thread-1 | FeaturesServiceImpl | 6 - org.apache.karaf.features.core - 4.0.9 | org.opendaylight.mdsal.model.ietf-lisp-address-types-2015-11-05/1.0.0.SNAPSHOT (Should be wired to: org.opendaylight.honeycomb.vbd.shared-models/1.6.0.SNAPSHOT (through [org.opendaylight.mdsal.model.ietf-lisp-address-types-2015-11-05/1.0.0.SNAPSHOT] osgi.wiring.package; filter:="(osgi.wiring.package=org.opendaylight.yang.gen.v1.urn.ietf.params.xml.ns.yang.ietf.lisp.address.types.rev151105)")) A believe that is a fault of Honeycomb/Vbd project, no action from Odlparent is needed. |
| Comment by Vratko Polak [ 16/Aug/17 ] |
|
> fault of Honeycomb/Vbd project Opened https://bugs.opendaylight.org/show_bug.cgi?id=8997 to track that. |