Resuming from hibernation crashes the application (Windows)

jdunkerley commented 1 month ago

On a laptop waking back up, the application reports the following error:

This was the JSON port of the language server: [org.enso.languageserver.boot.LanguageServerComponent] Started server at json:127.0.0.1:60829,

enso-bot[bot] commented 3 weeks ago

Adam Obuchowicz reports a new STANDUP for today (2024-10-31):

Progress: Fought much with setting up/updating the Windows setup. Checked that linux is fine. Discovered that it is LS who turns itself down after apparent disconnection from the client (and client crashes as it cannot connect to engine anymore). It should be finished by 2024-11-08.

Next Day: Next day I will be working on the #11389 task. Think and discuss how could be done with above.

hubertp commented 2 weeks ago

I will test it as well on some old WIndows machine

enso-bot[bot] commented 2 weeks ago

Adam Obuchowicz reports a new STANDUP for yesterday (2024-11-04):

Progress: As the Language Server seems to be closed immediately instead of waiting for further reconnections, I left that part to Hubert. Tried to add exception handling, so electron app at least won't crash It should be finished by 2024-11-08.

Next Day: Next day I will be working on the same task. Try to finish this task and take next one

hubertp commented 2 weeks ago

I can confirm. From the logs it looks like something requests a shutdown. Will investigate. For the moment unassigning GUI unless I confirm otherwise.

hubertp commented 2 weeks ago

This has been fixed a while ago, at least the backend part, in

7470
7729

So it appears to be a regression. On the other hand, it does not seem to trigger the delayed shutdown logic and it only happens on Windows, so could be a new bug.

enso-bot[bot] commented 2 weeks ago

Hubert Plociniczak reports a new STANDUP for yesterday (2024-11-05):

Progress: Adding more docs to #11265. Reproduced #11389 and investigating. It should be finished by 2024-11-08.

Next Day: Next day I will be working on the #11389 task. Address PR review, look into hibernation issues.

enso-bot[bot] commented 2 weeks ago

Hubert Plociniczak reports a new STANDUP for the provided date (2024-11-06):

Progress: Still investigating, looks like websockets aren't being closed properly in a regular case either (seeing lots of Peer did not acknowledged CLOSE frame messages). Probably Ydoc is at fault. It should be finished by 2024-11-08.

Next Day: Next day I will be working on the #11389 task. Continue investigating

enso-bot[bot] commented 1 week ago

Hubert Plociniczak reports a new STANDUP for the provided date (2024-11-07):

Progress: Something is triggering executor shutdown which then triggers LS shutdown. Hindered debugging due to lack of sufficient logging/missing dev setup on Windows. Tested polyglot version of NodeJS and situation is actually worse - suspend doesn't work on Linux either. Will file a ticket. It should be finished by 2024-11-08.

Next Day: Next day I will be working on the #11389 task. Continue investigating

hubertp commented 1 week ago

Tested polyglot version of NodeJS

That obviously meant polyglot version of Ydoc

enso-bot[bot] commented 1 week ago

Hubert Plociniczak reports a new STANDUP for the provided date (2024-11-08):

Progress: Continue investigating for the source of the bug. Will use PR artifacts to build and test suspend on Windows with additional logging. Filed #11515 for additional problems with Graal-based Ydoc. As a distraction, fixed #11262, no idea how to unit test it though. It should be finished by 2024-11-08.

Next Day: Next day I will be working on the #11389 task. Continue investigating

enso-bot[bot] commented 1 week ago

Hubert Plociniczak reports a new STANDUP for yesterday (2024-11-12):

Progress: Firefighting CI. While not doing that, analyzing logs for suspend. The issue appears to be rather deep in the stack, hence delays. It should be finished by 2024-11-15.