Rules refuse to load at startup due to jars in addons folder #3676
Labels
bug
An unexpected problem or unintended behavior of the Core
Comments
morph166955
added
the
bug
|
This issue has been mentioned on openHAB Community. There might be relevant details there: https://community.openhab.org/t/openhab-4-0-snapshot-discussion/142322/387 |
|
After installing the snapshot 3519 with almost nothing setup except few items, one DSL rule and one sitemap, I encountered a case where my feeling was that the rule was no more executed. My scenario was very different from yours (I updated Basic UI with bundle:update) but the result was apparently the same. So there may be a severe bug with rules recently introduced. I will have to investigate more to provide a reproducible scenario and confirm that the problem is really about rules. At the same time, I upgraded my production openHAB server to that version without any problem. |
|
Can you please check if |
|
I'm not near it today, but can probably pull logs tomorrow. |
|
This issue has been mentioned on openHAB Community. There might be relevant details there: https://community.openhab.org/t/openhab-4-0-milestone-discussion/145133/490 |
|
Same Problem here, I don't have any addons in the addon folder: |
|
Is there any bundle not "Active" when you list them? |
|
Yes: |
|
If you uninstall them (instead of stopping them), do you have the same issue? |
|
No as soon as all bundles are "Active" OH 4.0.0M4 starts up normally. |
|
Lots o logs... I tried to run date/time when commands executed to line it up to the logs. System started at 17:10:22.188 System was basically loaded at 17:10:39.980 (Everything was Active by this point including the 4x jar addons as seen below): 296 │ Active │ 80 │ 4.0.0.202305151551 │ org.openhab.binding.ring At 17:15:50 I moved the rachio jar out of the addons folder At 17:15:59.091 system notes All automation bundles ready. At 17:16:07.833 system notes Rule engine started. I reran the test by moving the androidtv jar out instead of rachio, same effect. androidtv is merged in at this point, that jar is almost identical to what's merged, it just includes the two current pending PRs which shouldn't have any impact. |
|
I think I know the cause of this issue. Is anyone able to reproduce it for sure? I can provide a bundle that should fix it. |
If I stop one of the bundles in console, the Rule Engine doesn't start and event.log stays empty... |
It happens every start for me. Easy enough to recreate. |
|
Install the fixed bundle with
and restart. Is everything working now? |
|
So a weird thing happened when I tried it. The upgrade worked: When I restarted, it was back to the original version (all I did was turn the docker off and back on, I didn't clear any caches). However, when it came up it loaded like it was going to jam, then about 30 seconds later it sent: And then it loaded my rules for a second time. At that point, they started working. |
|
That's really strange. Does it happen again when you restart again? Which bundle is now shown by |
|
It shows the original that came with the image. 4.0.0.202306251610. Maybe it's a docker thing? |
|
Maybe. I don't use docker. Always trouble with things like that. |
|
Is there a cleaner way to reload/restart from inside karaf? Any chance it's because these are sitting at Waiting? openhab> bundle:list -s | grep script |
|
No, they are waiting because they depend on the update bundle. This is why you need to restart. But of course, if everything is reverted, it'll not work. |
|
I was able to catch it early on the restart. It looks like it dumped the update... openhab> bundle:list -s | grep script |
|
Maybe push it and I'll just update the snapshot when it runs? |
|
@lolodomo Since you seem to experience the same issue: Can you try if the fix resolves the start level 20 issue for you? |
|
Yes, I plan to test this evening. |
|
@J-N-K : unfortunately, it does not solve the start level 20 I encounter in the issue #3680 with a OH server running in Windows 10. I updated the bundle org.openhab.core.automation.module.script with the one compiled from your PR #3683: |
|
Did you restart? |
|
@J-N-K : I added a log at the beginning of the checkReady method: And the result is: Looks like my start level is 10 when this method is called while your code is now expecting 20. |
Of course. |
|
So previous code with StartLevelService.STARTLEVEL_OSGI was probably better in my case ? |
|
Or |
|
Level 20 should be correct, I have added some more logging, maybe you can try to update again. In principle, the ready service should notify the tracker that start level 20 was reached. I also added logging for that. |
I don't find any change in your branch. Did you really push a new commit ? |
|
I only updated the download link. Changes pushed. |
|
With your last changes: |
|
With TRACE enabled: |
|
Uh. That's a neat one. The start-level service still returns 10 when |
|
Updated once again. |
|
Now it works, start level is 100. |
|
Snapshot 3528 looks good! Thank you! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Upgraded from 3485 to 3519. I have 4 jars in my addons folder, and have for some time now. Since upgrading, at start the rules never fire. I let it sit for over 15 minutes at one point. Thinking I had a bad jar (which was unlikely as they all worked on 3485) I moved them out of the addons folder and immediately saw:
2023-06-30 15:03:00.153 [INFO ] [nal.ScriptEngineFactoryBundleTracker] - All automation bundles ready.
After further testing, moving ANY jar out of the addons folder seems to clear out the issue. It is irrelevant which jar is moved, it seems that the identification that a change happened is enough to break it free. I can move the jar right back in and everything works as expected.
The text was updated successfully, but these errors were encountered: