Navigating from an unsaved logstash pipeline causes a Kibana fatal uncaught error

[liza-mae]

Kibana version: 6.3.0 BC7
Elasticsearch version: 6.3.0 BC7

Server OS version: Ubuntu tar

Browser version: Firefox / Chrome Latest

Browser OS version: OSX

Original install method (e.g. download page, yum, from source, etc.):
From staging

Description of the problem including expected versus actual behavior:

When you are editing a logstash pipeline and decide to navigate from the page without saving, I am getting a fatal error. I see 1000+ repeating messages in the console. This is not happening in 6.2.4 and I was not able to reproduce it in 6.3.0 BC6. In the working cases: a prompt comes up asking to either discard changes or cancel.

From pipeline

TypeError: Cannot read property '$invalid' of undefined
at PipelineEditController.get (kibana.bundle.js:7)
at fn (eval at compile (vendors.bundle.js:133), :4:269)
at Scope.$digest (vendors.bundle.js:133)
at vendors.bundle.js:133
at completeOutstandingRequest (vendors.bundle.js:133)
at vendors.bundle.js:133
(anonymous) @ vendors.bundle.js:133
(anonymous) @ vendors.bundle.js:133
$digest @ vendors.bundle.js:133
(anonymous) @ vendors.bundle.js:133
completeOutstandingRequest @ vendors.bundle.js:133
(anonymous) @ vendors.bundle.js:133
setTimeout (async)
Browser.self.defer @ vendors.bundle.js:133
$evalAsync @ vendors.bundle.js:133
hideTooltipBind @ commons.bundle.js:3
dispatch @ vendors.bundle.js:128
elemData.handle @ vendors.bundle.js:128
commons.bundle.js:3 Error: Uncaught Error: [$rootScope:infdig] 10 $digest() iterations reached. Aborting!
Watchers fired in the last 5 iterations: []
http://errors.angularjs.org/1.6.9/$rootScope/infdig?p0=10&p1=%5B%5D (https://localhost:5607/bundles/vendors.bundle.js:133)
at window.onerror (commons.bundle.js:3)
fatalError @ commons.bundle.js:3
window.onerror @ commons.bundle.js:3
error (async)
(anonymous) @ commons.bundle.js:3
webpack_require @ vendors.bundle.js:3
(anonymous) @ commons.bundle.js:3
webpack_require @ vendors.bundle.js:3
(anonymous) @ commons.bundle.js:3
webpack_require @ vendors.bundle.js:3
(anonymous) @ commons.bundle.js:3
webpack_require @ vendors.bundle.js:3
(anonymous) @ commons.bundle.js:3
webpack_require @ vendors.bundle.js:3
(anonymous) @ commons.bundle.js:3
webpack_require @ vendors.bundle.js:3
(anonymous) @ commons.bundle.js:3
webpack_require @ vendors.bundle.js:3
(anonymous) @ commons.bundle.js:3
webpack_require @ vendors.bundle.js:3
(anonymous) @ commons.bundle.js:3
webpack_require @ vendors.bundle.js:3
5127 @ kibana.bundle.js:3
webpack_require @ vendors.bundle.js:3
window.webpackJsonp @ vendors.bundle.js:3
(anonymous) @ kibana.bundle.js:3
vendors.bundle.js:133 Uncaught Error: [$rootScope:infdig] 10 $digest() iterations reached. Aborting!
Watchers fired in the last 5 iterations: []
http://errors.angularjs.org/1.6.9/$rootScope/infdig?p0=10&p1=%5B%5D
at vendors.bundle.js:133
at Scope.$digest (vendors.bundle.js:133)
at vendors.bundle.js:133
at completeOutstandingRequest (vendors.bundle.js:133)
at vendors.bundle.js:133
(anonymous) @ vendors.bundle.js:133
$digest @ vendors.bundle.js:133
(anonymous) @ vendors.bundle.js:133
completeOutstandingRequest @ vendors.bundle.js:133
(anonymous) @ vendors.bundle.js:133
setTimeout (async)
Browser.self.defer @ vendors.bundle.js:133
$evalAsync @ vendors.bundle.js:133
hideTooltipBind @ commons.bundle.js:3
dispatch @ vendors.bundle.js:128
elemData.handle @ vendors.bundle.js:128
Steps to reproduce:

Install Elasticsearch, Kibana
From logstash pipeline
Create a new pipeline
Fill out info and then navigate to Discover, the fatal error occurs. It also occurs if you edit an existing pipeline, don't save and navigate out. It is expected behavior to get a prompt to either discard changes or cancel the navigation.

[liza-mae]

Looks like we might workaround this by disabling the prompt for 6.3.0 and narrow down a root cause and fix will be referenced here #19594.

[justinkambic]

As a note to the process of reproducing this issue - it wasn't happening for me until I had created an index with some data and created an index pattern. After I had done that, I navigated from Discover to Pipeline Management and was able to reproduce the issue as described here.