Updating to address @rexledesma's comments

lgtm besides the comment

lgtm

Nice!

Lgtm! Nice find w/ pytest-rerun

Based on discussion with @rexledesma, refactoring to just use Container as a schema

Allowing for imagePullPolicy as well

Sounds good to me! This would break run attribution in the demo, right, since that tries to read the header (as a proof of concept for the prezi ask, I think?)

Addressing 'TODO'

responding to @rexledesma's comments

requiring config of helm_dir_path & subchart_paths via @rexledesma's comments

Also increase number of retires on helm checks since that appears to be a problem

more docstring edits

Fix docstrings

This looks fine to me. I'll take a look at that Drizly problem ASAP.

lgtm! Haven't run into the Interface error personally but I figure if it's working locally for you it's fine!

responding to @rexledesma's comments

Removing patch on warnings

In D8043#214316, @rexledesma wrote:

This example needs to show up in the docs somewhere, probably in https://docs.dagster.io/deployment/run-coordinator. Similar to https://docs.dagster.io/deployment/guides/docker#example, we should highlight the fact that a custom run coordinator can be used, and that an example is a run coordinator that injects run tags

rebasing & prefixing test class with Test** to adhere to pytest conventions

prefixing with Test to adhere to pytest conventions

rebasing

Renaming to QueuedRunCoordinatorTests as per @jordansanders' advice

Fix tox

Fix import

rename run_attribution -> run_attribution_example

restructuring to address @rexledesma's comments as well as rebasing

This issue was there to track a mypy problem because we had a pin in SqlAlchemy for <1.4 (I think related to this diff). I'll add in the type annotation in a separate diff - this LGTM. thanks for catching this Alex.

Oops, looks like I missed this - I'll work on getting these under test ASAP.

Adding in asserts & cleaning up tests

Added tests & removed the NullPool wait_timeout fix - there's no pool for that case so there's nothing to recycle

maybe fix lint before you push? otw lgtm.

Renaming to Dagster-Session-Token

rebasing & addressing comments

Oops, sorry for missing this! thanks for the fix.

Adding comment and renaming to headerAuthToken

In D7810#204948, @bengotow wrote:

Hmm I think this works for now because we're planning on using the authToken primarily to attribute runs, etc. to different users, but I think if we really wanted security we'd need to add auth to the websockets interface as well (or stop using it), right? I wonder if we should add a small comment in here explaining that there are still un-authorized GraphQL requests being made from the app in the few places we need websockets?

adding comment

Removing React.useMemo on rootServerURI as per @dish's suggestion

The removal of the GEventExecutor

I am pretty sure this is fine to do - and is effectively replicating the way the all-websocket approach worked - but we should add a test cases section specifically about making sure htttp requests can go through while the active run event log subscription is firing data back. You can do this by navigating around while another tab is actively watching a run of many_events from the toyes repo or something.

No longer overrwriting runs page on network errors

In D7688#203536, @alangenfeld wrote:

This is behavior that is due to the difference between websockets & http - the websocket response does not have an 'error' field on the JS side so the components render differently. Not sure if throwing away the error in this case (to conform to the old behavior) makes sense though -- feels like errors are important to pay attention to!

Hmm - in the screenshot shown we show "GraphQLError - see console for details" but the issue is just that there is no connectivity right? I personally think thats a rough regression. Replacing the contents of the page a big error warning when a polling-refresh query fails due to network connectivity seems bad to me.

Im not sure the exact fix - but I do think we should distinguish "the server sent back an error because something bad happened" from "you couldn't reach the server"

Removing automation change

Updating changelog

Reformatting reload_repository_location docs