Tests flaking due to `websocket url timeout reached`

[uhoreg]

In matrix-org/matrix-js-sdk#4566, the complement crypto test failed twice. It looks like it fails on the first test being run, timing out on initialization. So a timeout probably needs to be increased somewhere.

Output of the failure:

2024/12/04 11:41:33 Sharing [SERVER_NAME=hs1 SYNAPSE_COMPLEMENT_DATABASE=sqlite] host environment variables with container
  2024/12/04 11:41:40 Sharing [SERVER_NAME=hs2 SYNAPSE_COMPLEMENT_DATABASE=sqlite] host environment variables with container
  
  **********************************************************************************************
  Ryuk has been disabled for the current execution. This can cause unexpected behavior in your environment.
  More on this: https://golang.testcontainers.org/features/garbage_collector/
  **********************************************************************************************
  2024/12/04 11:41:46 github.com/testcontainers/testcontainers-go - Connected to docker: 
    Server Version: 26.1.3
    API Version: 1.45
    Operating System: Ubuntu 22.04.5 LTS
    Total Memory: 15981 MB
    Resolved Docker Host: unix:///var/run/docker.sock
    Resolved Docker Socket Path: /var/run/docker.sock
    Test SessionID: 68f4e740a8b5bc1b66c578875a8351e0820b0aa00d63f66e510201081bf77751
    Test ProcessID: 566213f1-3462-4f54-9753-919dca62ac31
  2024/12/04 11:41:46 🐳 Creating container for image mitmproxy/mitmproxy:10.1.5
  2024/12/04 11:41:46 ✅ Container created: 2224e487dd3b
  2024/12/04 11:41:46 🐳 Starting container: 2224e487dd3b
  2024/12/04 11:41:46 ✅ Container started: 2224e487dd3b
  2024/12/04 11:41:46 🚧 Waiting for container id 2224e487dd3b image: mitmproxy/mitmproxy:10.1.5. Waiting for: &{timeout:<nil> Log:loading complement crypto addons IsRegexp:false Occurrence:1 PollInterval:100ms}
  2024/12/04 11:41:50 🔔 Container is ready: 2224e487dd3b
      deploy.go:205: ComplementCryptoDeployment created (network=complement_crypto_dirty):
      deploy.go:206:   NAME          INT          EXT
      deploy.go:207:   synapse:      hs1          http://127.0.0.1:32769/ (rp=http://127.0.0.1:32777)
      deploy.go:208:   synapse:      hs2          http://127.0.0.1:32774/ (rp=http://127.0.0.1:32776)
      deploy.go:209:   mitmproxy:    mitmproxy    controller=http://127.0.0.1:32775/
      client.go:685: [CSAPI] POST hs1/_matrix/client/v3/register => 200 OK (20.990225ms)
      client.go:685: [CSAPI] POST hs1/_matrix/client/v3/register => 200 OK (16.981548ms)
      client.go:685: [CSAPI] POST hs1/_matrix/client/v3/createRoom => 200 OK (99.561305ms)
  JS SDK listening on http://127.0.0.1:45249/
      lang_jssdk.go:42: must.NotError: NewJSClient: %s -> failed to RunHeadless: failed to navigate to http://127.0.0.1:45249: websocket url timeout reached

[richvdh]

https://github.com/matrix-org/complement-crypto/actions/runs/18689875867/job/53292486368 flaked similarly

[richvdh]

This seems to be happening more often. I've noticed it seems to particularly affect TestDelayedInviteResponse/js (46.89s), though I think that's just because that's the first test that gets run.

[richvdh]

I've added some more logging in #232, which I hope will provide elucidation next time this happens