Port XXXX Not Found -- The constant need to clear gitpod cookies

[smacintyre]

Bug description

The past couple months my team has been seeing constant "Port XXXX Not Found" errors when opening previews in the browser. The application is running and the port is being served; the work around is clearing all gitpod.io cookies and stored data and reopening the workspace then it will be served.

• The majority of the it works, but this is happening 2-3 per team member per week, it adds up
• Clearing gitpod.io cookies always resolves the problem
• Often, not always, it happens the first time a workspace is open on a new cluster: i.e., the first time opening on ws-us31.gitpod.io -- it's pretty much 100% when a new cluster is opened

Steps to reproduce

• Use gitpod regularly
• Open a workspace on a new cluster
• Open a port
• Attempt to browse to it

Workspace affected

No response

Expected behavior

I should see the content served at the port

Example repository

No response

Anything else?

No response

[sagor999]

@smacintyre would you be able to open developer console in your browser when this happens and copy paste errors from it in here?
That will help to track this down.
I tried to repro this locally but had no luck so far.

[smacintyre]

@sagor999 Will do. It happens really randomly.

[kylos101]

Hey @smacintyre we're switching to a new cluster now, over the next 3 hours it'll be come more likely that you'll land on ws-us34.gitpod.io. If you encounter trouble, and are able to share any errors, that'd be great!

[smacintyre]

@sagor999 @kylos101 Thank you for your response. Unfortunately (fortunately?) the problem has not reoccurred and there were no issues with the move to ws-us34. So I don't have any more feedback. It seems since I opened this issue it has largely stopped. I'm sorry I haven't been able to provide any more details. Thank you for your help. If you wish to close this ticket, please go ahead.

Oh, any word on when the promised Asian zone will be online?

[kylos101]

Hey @smacintyre okay, we'll close this issue, but, you're welcome to reopen if it rears its head again. 🙂

You may wish to subscribe to this issue. I cannot share an ETA now, but, if you subscribe to it that help you keep track of the related conversation.