Warning message ( Page Not Found ) seen on the UI as soon as you navigate to ML app- also seen in Discover app.

[rashmivkulkarni]

Kibana - BC6 of 8.4.0

License Level: Basic

When I navigated from Discover to Machine Learning app, I saw the warning which was appearing on the app as seen in the screenshot below.

Nothing much was captured in the browser console except

Logs were clean.

@LeeDr hit a similar issue on Discover app as well.

This doesnt seem limited to ML alone . Deeper investigation needed.

cc @qn895 @LeeDr

[rashmivkulkarni]

@elastic/kibana-data-discovery @elastic/kibana-vis-editors looping the teams in for more inputs.

FYI: didn't see it on BC6 of 8.4.0 on cloud .

more chatter happened here: https://elastic.slack.com/archives/C5TQ33ND8/p1661189973080439

[elasticmachine]

Pinging @elastic/kibana-vis-editors @elastic/kibana-data-discovery

[LeeDr]

I just reproduced the issue in Discover on BC2 build of 8.4.0

My steps to reproduce;

1. I'm running the stack installed on Rocky 8 linux installed via rpms with security enabled with TLS and Trial license
2. Go to Stack Monitoring.
3. Go to Alerts and Rules in the top right corner and select Manage rules (note that this page has a 5m auto-refresh interval built-in)
4. Using the Nav bar to navigate to Discover
5. open the timepicker and enable auto-refresh for a few second interval
6. The message appears on the page

[LeeDr]

I reproduced it in Discover on BC1 as well

[flash1293]

Doesn’t seem related to vis editors. Data visualizer is owned by the ml team, Discover is owned by the data discovery team

[elasticmachine]

Pinging @elastic/kibana-core (Team:Core)

[rashmivkulkarni]

This is a front-end issue, which happens when there is a refresh interval set in an app and you navigate away to a different app - You do see the warning message appearing.

[droberts195]

This looks the same as #137180.

I also saw something similar once, but it wasn’t consistently reproducible at the time. It looked like there was a # in the wrong place in the URL in a Slack discussion at the time.

[rudolf]

Seems like it was introduced in 8.3.0 #133157

[lukeelmers]

Worth noting that ML & Discover each have their own implementations of this warning, so if we are seeing it in both places, I'm wondering if there's something else going on.

It'd be good to get @elastic/kibana-data-discovery's eyes on this.

[lukeelmers]

We should consider documenting this as a known issue in 8.4.0, and push to get it resolved in 8.4.1.

It sounds like the workarounds would be either:

• disable autorefresh; or
• refresh your browser page

Do those workarounds sound right @rashmivkulkarni?

[rashmivkulkarni]

Yes, its good to document this as a known issue , for 8.4.0 and fix it for 8.4.1.

The work around is right . The warning message does go away upon page refresh and/or disable auto-refresh.

[kertal]

I could reproduce in 8.4 and in a main based PR. Both Discover and ML have the same problem when navigating from stack monitoring, have a look:

discover-ml-timepicker-troubles.mp4

What is weird here in the Discover case is, that the page that cannot be found, is the route of the stack monitoring plugin ... no-data. It could be an underlying problem of a service both plugins consume. So the time picker is syncing to the URL, maybe here's the problem? FYI @elastic/kibana-app-services maybe @Dosant could take a look at the URL syncing (I did but I think it needs more 👀 on this problem)

[kertal]

Also reproduced using Lens

timepicker-lens-404.mp4

[rashmivkulkarni]

I get this error when I add sample data and navigate to dashboard from the sample data drop menu . Its transient .

Page not found

Dashboard application doesn't recognize this route: /tutorial_directory/sampleData.

[Dosant]

FYI https://github.com/orgs/elastic/teams/kibana-app-services maybe @Dosant could take a look at the URL syncing (I did but I think it needs more 👀 on this problem)

@kertal, good call.

Here is the class that monitoring is using for URL syncing: https://github.com/elastic/kibana/blob/main/x-pack/plugins/monitoring/public/url_state.ts and I couldn't find if destroy() is ever called

kibana/x-pack/plugins/monitoring/public/url_state.ts

Lines 141 to 145 in 1a70f6f

	public destroy() {
	this.syncQueryStateWithUrlManager.stop();
	this.stateContainerChangeSub.unsubscribe();
	this.stateSyncRef.stop();
	}

I put debugger into it and didn't see it triggered when navigating from stack monitoring. Also didn't find it called in the code.

I believe this could cause any kind of weirdness with the URL because after you navigate to monitoring, the code keeps listening to time picker updates and is trying to update the URL that doesn't belong to the app.

I believe the team should look into fixing this @elastic/stack-monitoring, but I don't think this is related to any other bugs that are reproduced when you don't navigate to monitoring first

Also related: #135136

[Dosant]

Regarding the issue from the description with Basic + ML:

When I navigated from Discover to Machine Learning app, I saw the warning which was appearing on the app as seen in the screenshot below.

I agree with @rudolf #139231 (comment), as ml seems to be looping itself because of this redirect:

kibana/x-pack/plugins/ml/public/application/license/ml_client_license.ts

Lines 49 to 52 in 1ae09bc

	function redirectToBasic() {
	window.location.href = '#/datavisualizer';
	return Promise.reject();
	}

cc @darnautov

[darnautov]

@Dosant this redirect in the licence check has been introduced in #59275 (7.7.0).

@rudolf pointed to the PR that introduced a redirect from unknown routes with Redirect from react-router-dom (8.3.0)

[Dosant]

@Dosant this redirect in the licence check has been introduced in #59275 (7.7.0).

@rudolf pointed to the PR that introduced a redirect from unknown routes with Redirect from react-router-dom (8.3.0)

@darnautov, Could it be that the basic license redirect wasn't properly tested with #133157?
What concerns me is a hashbang in window.location.href = '#/datavisualizer'; and maybe it causes to trigger "unknown" route

[darnautov]

Created an issue #139279, aiming to fix for 8.4.1.

[flash1293]

Seems like this is fixed, so this issue can be closed, right @darnautov ?

[kertal]

I think the issue of stack monitoring has not been resolved? could you have a look dear @elastic/stack-monitoring team?

[elasticmachine]

Pinging @elastic/infra-monitoring-ui (Team:Infra Monitoring UI)

[darnautov]

@flash1293 @kertal indeed I only fixed the ML app, I assume the stack monitoring team might consider fixing it for 8.4.2.

[smith]

@kertal I tried to reproduce on stack monitoring running locally on the latest main. Not seeing it happen but I could be doing it wrong.

[kertal]

@smith Just testing the logs explorer PR, tried to reproduce the issue, have a look:

Kapture.2022-08-29.at.22.20.01.mp4

[smith]

@kertal looks like we've previously had this reported in #135136. We'll get it on our board to be picked up.

[Kerry350]

I've opened #139908 to fix the Stack Monitoring side.