Blank or white Admin page, Rooms and Spoke are fine

[aidanhmiles]

Hello. I previously had a working self-hosted Hub, and I'm sure I broke it by messing with CORS and CSP. At first, we couldn't open rooms, spoke, or the admin, then I tried both restoring and deleting CORS and CSP params from the AWS parameter store (to try to revert to default values), and I rolled over all instances after each config change.

At this point, I can create Rooms and use Spoke, but cannot access the Admin interface (I see a white, blank page). Since the Hubs team is moving over to discussions and I see no previous discussion of the white/blank admin page issue, I wanted to open one in the hopes of cataloguing reasons that anyone might experience this issue.

Other observations:

• Admin page HTML source shows an empty #ui-root element
• Admin page HTML source lacks env:* or ret:* meta tags
• Spoke and Room pages have env:* or ret:* meta tags
• No network requests appear to fail on any page other than favicon

• All pages show the same three warnings, 1) cookies and the sameSite atttribute, and 2) trying to connect to an empty hostname from something, both over http and ws. I believe this is new behavior, and suggests to me that a value is missing for a hostname config somewhere. I have no idea where that might be.

• I've cleared site data multiple times throughout my testing, just in case

• The only params I have left in the parameter store are:
  /ita/{hubname}/hubs/deploy/type
  /ita/{hubname}/reticulum/resolver/google_poly_api_key
  /ita/{hubname}/reticulum/resolver/sketchfab_api_key
  /keymaster/{hubname}/guardian-secret-key
  /keymaster/{hubname}/janus-and-reticulum-admin
  /keymaster/{hubname}/phoenix-key
  /keymaster/{hubname}/postgrest-db-secret
  /keymaster/{hubname}/reticulum-bot-access-key
  /keymaster/{hubname}/reticulum-cookie
  /keymaster/{hubname}/reticulum-oauth-token
  /keymaster/{hubname}/send-email-password

Let me know if I can provide more details or anything. Any ideas, tips, questions, etc are most appreciated.

[robinkwilson]

Thank you for the detailed description of what you're seeing and what you've tried. Helps a lot!

Do you remember the CSP parameter(s) you deleted?
Did you deploy Cloudflare workers before deleting the parameters in the parameter store?

[robinkwilson]

You have a custom client!

Run npm run undeploy to remove the custom client and revert back to the HC default client. Let's see what happens!

[robinkwilson]

Also fetch from the hubs-cloud branch in the hubs repo to ensure you have the latest deployed HC client version if you're going to try deploying the custom client again.

[aidanhmiles]

I believe we incorporated the latest from your repo last week. I'll double-check.

There's some (working) code in the custom client that we need for some meetings today, but I'll undeploy and report back this evening. Thank you!

[robinkwilson]

Marking as answer for now to get this out of our todo pipeline. Please @ me to reopen.

[aidanhmiles]

I can confirm, npm run undeploy fixed whatever the issue was. Still unsure what happened. I'll compare our custom client to master and see if anything useful pops out. Thanks again for your help!