Ensure that the Mediaflux Status indicator functions for non-local deployments

[jrgriffiniii]

Currently, the WebSockets/ActionCable feature providing the Mediaflux API connectivity status does not work for non-local deployments of the application (i. e. staging, qa, and production). This is going to remain blocked until modifications are provided for the NGINX+ proxy permitting WebSockets traffic for the host server nodes within each deployment environment.

Acceptance Criteria:

• [ ] Deploy into the staging environment and ensure that the status indicator is behaving as expected
• [ ] Deploy into the qa environment and ensure that the status indicator is behaving as expected
• [ ] Deploy into the production environment and ensure that the status indicator is behaving as expected

[jrgriffiniii]

https://github.com/pulibrary/tiger-data-app/tree/jrgriffiniii-fix-mflux and https://github.com/pulibrary/princeton_ansible/tree/jrgriffiniii-nginx-websockets-tigerdata effectively addressed this for the staging environment.

[jrgriffiniii]

The following is being logged following a redeployment to staging:

I, [2024-07-22T14:05:02.571000 #3047744]  INFO -- : [95eef1d8-b6a7-4333-8ca3-663721f5f41d] [dd.env=staging dd.service=tigerdata dd.version=1.0.0 dd.trace_id=1125728829946400298 dd.span_id=869445013784936239 ddsource=ruby] Started GET "/cable"[non-WebSocket] for 172.20.80.14 at 2024-07-22 14:05:02 +0000
E, [2024-07-22T14:05:02.571291 #3047744] ERROR -- : [95eef1d8-b6a7-4333-8ca3-663721f5f41d] [dd.env=staging dd.service=tigerdata dd.version=1.0.0 dd.trace_id=1125728829946400298 dd.span_id=869445013784936239 ddsource=ruby] Failed to upgrade to WebSocket (REQUEST_METHOD: GET, HTTP_CONNECTION: , HTTP_UPGRADE: )
I, [2024-07-22T14:05:02.571529 #3047744]  INFO -- : [95eef1d8-b6a7-4333-8ca3-663721f5f41d] [dd.env=staging dd.service=tigerdata dd.version=1.0.0 dd.trace_id=1125728829946400298 dd.span_id=869445013784936239 ddsource=ruby] Finished "/cable"[non-WebSocket] for 172.20.80.14 at 2024-07-22 14:05:02 +0000

[jrgriffiniii]

Using cURL, the following does properly resolve for the passenger-managed deployment of the Rails app:

curl -v -i -N -H "Connection: Upgrade" -H "Upgrade: websocket" -H "Host: localhost" -H "Origin: localhost"  -H "X-Passenger-Connect-Password: [...]" http://0.0.0.0:44873/cable/

[jrgriffiniii]

I can now verify that only the NGINX configuration on the load balancer must be adjusted:

$ curl -v -i -N -H "Connection: Upgrade" -H "Upgrade: websocket" -H "Host: tigerdata-staging.lib.princeton.edu" -H "Origin: https://tigerdata-staging.lib.princeton.edu" http://0.0.0.0/cable/

> GET /cable/ HTTP/1.1
> Host: tigerdata-staging.lib.princeton.edu
> User-Agent: curl/7.81.0
> Accept: */*
> Connection: Upgrade
> Upgrade: websocket
> Origin: https://tigerdata-staging.lib.princeton.edu
>
* Mark bundle as not supporting multiuse
< HTTP/1.1 101 Switching Protocols
HTTP/1.1 101 Switching Protocols
< Connection: upgrade
Connection: upgrade
< Status: 101 Switching Protocols
Status: 101 Switching Protocols
< Upgrade: WebSocket
Upgrade: WebSocket
< WebSocket-Origin: https://tigerdata-staging.lib.princeton.edu
WebSocket-Origin: https://tigerdata-staging.lib.princeton.edu
< WebSocket-Location: wss://tigerdata-staging.lib.princeton.edu/cable/
WebSocket-Location: wss://tigerdata-staging.lib.princeton.edu/cable/
< Date: Tue, 23 Jul 2024 14:45:13 GMT
Date: Tue, 23 Jul 2024 14:45:13 GMT
< X-Powered-By: Phusion Passenger(R) 6.0.22
X-Powered-By: Phusion Passenger(R) 6.0.22
< Server: nginx/1.18.0 + Phusion Passenger(R) 6.0.22
Server: nginx/1.18.0 + Phusion Passenger(R) 6.0.22

[jrgriffiniii]

The following is behaving as expected:

wscat --origin=https://tigerdata-staging.lib.princeton.edu -L -c ws://tigerdata-staging2.lib.princeton.edu/cable/
Connected (press CTRL+C to quit)
< {"type":"welcome"}
< {"type":"ping","message":1721756269}
[...]
wscat --origin=https://tigerdata-staging.lib.princeton.edu -L -c ws://tigerdata-staging1.lib.princeton.edu/cable/
Connected (press CTRL+C to quit)
< {"type":"welcome"}
> %