[MGNLUI-6736] Frozen admincentral after restart (after major vaadin update) Created: 09/Jun/21  Updated: 18/Oct/22  Resolved: 06/Aug/21

Status: Closed
Project: Magnolia UI
Component/s: None
Affects Version/s: 6.2.7
Fix Version/s: 6.2.12

Type: Bug Priority: Neutral
Reporter: Roman Kovařík Assignee: Roman Kovařík
Resolution: Fixed Votes: 1
Labels: VN-Testing, maintenance
Remaining Estimate: Not Specified
Time Spent: 7h
Original Estimate: Not Specified

Attachments: File AssetsCoreFunctionalTests_addFolderToRoot.flv.mp4    
Issue Links:
Problem/Incident
causes MGNLUI-6871 Grid row not updating reliably after ... Closed
is caused by BUILD-441 Update Vaadin to version 8.12.4 Closed
Relates
dependency
documentation
to be documented by MGNLUI-6737 DOC: Admincentral freezes after Vaadi... Closed
relation
is related to BUILD-478 Update Vaadin to latest 8.13.x Closed
is related to MGNLUI-6810 DOC: Remove Admincentral freeze after... Closed
supersession
is superseded by MGNLUI-6899 Jetty libraries on the classpath make... Closed
Template:
Acceptance criteria:
Empty
Task DoD:
[X]* Doc/release notes changes? Comment present?
[X]* Downstream builds green?
[X]* Solution information and context easily available?
[X]* Tests
[X]* FixVersion filled and not yet released
[ ]  Architecture Decision Record (ADR)
[X]* (Converted to MGNLUI-6810) Remove from known issues docu [Ashraf Khamis]
Bug DoR:
[X]* Steps to reproduce, expected, and actual results filled
[X]* Affected version filled
[X]* Verify with 8.13.1
Release notes required:
Yes
Documentation update required:
Yes
Date of First Response:
Epic Link: Vaadin push issues

 Description   

Steps to reproduce

  1.  Start CE demo webapp
  2.  Log in into admincentral
  3. Interact with UI (e.g. click a find bar result)

Expected results

UI responds.

Actual results

Progress bar is pending infinitely (most of the times, although this should be tested repetitively).

Workaround

Refresh the browser tab.

Development notes

This is caused by UI#push calls before the push connection is established (in our case findbar results).
Although vaadin has code to handle these pending pushed, this causes the freezing.
https://github.com/vaadin/framework/blob/e30226a19564c6ee77064e5178a7665554d57466/server/src/main/java/com/vaadin/server/communication/AtmospherePushConnection.java#L256 (commenting out this line fixes the issue)



 Comments   
Comment by Christoph Meier [ 08/Aug/21 ]

fgrilli - It is indeed possible that this issue is responsible for some of the "arbitrary" failures - which very often (if not always) appear on AssetsCoreFunctionalTests.
I will have a look at this in the coming week.
Thanks for the reminder.

Comment by Christoph Meier [ 11/Aug/21 ]

fgrilli - F.Y.I.
Yesterday I was running PagesCoreFunctionalTests and AssetsCoreFunctionalTests many times (locally).

There is still a high chance that the very first test, which is run, fails. Especially after I did restart the webapp.
If I run AssetsCoreFunctionalTests, on my env, #addSubFolders is always executed first, and it always fails, if that test run was just after the webapp started.
It fails on L.89

expect.findBar().openApp("Assets");

When I then re-run the #addSubFolders with many repetitions, but without restarting the webapp, the method passes always.

I'm not really sure how to interpret this - but at least it indicates that something is still fishy. If we tackle all the tickets of type bug in the test-fwk - we hopefully can get rid of the "arbitrary" failures.

Comment by Roman Kovařík [ 11/Aug/21 ]

cmeier could you try the same with 6.2.6 webapp?

Generated at Mon Feb 12 09:39:30 CET 2024 using Jira 9.4.2#940002-sha1:46d1a51de284217efdcb32434eab47a99af2938b.