TestSyncBlocksDuringPartialStateJoin test fails in Complement when using workers

[reivilibre]

This sounds an awfully lot like the faster room join stuff, which as far as I know is not implemented for use on workers at the moment.

It sounds like we want to skip the test in a workerised setup, but I'm not as of yet sure how to do that (nor am I absolutely certain that this is indeed what we want).

Interesting piece of log:

2022/05/20 16:47:21 fed.alice.hs1: Server is responding after 85 iterations
2022/05/20 16:47:21 fed.alice.hs1 -> http://localhost:49286 (26c7ede363b05b4cd004f2c9b211ea67ffca5508deb74cde6726603859ff1145)
    federation_room_join_partial_state_test.go:33: Deploy times: 13.805967618s blueprints, 5.127144386s containers
    federation_room_join_partial_state_test.go:55: Making GET request to http://localhost:49286/_matrix/client/r0/capabilities (syt_YWxpY2U_oGHIbbKayEgVymeIvaNI_0QUzop)
    client.go:544: GET hs1/_matrix/client/r0/capabilities => 200 OK (4.421158ms)
    federation_room_join_partial_state_test.go:55: HTTP/1.1 200 OK
        Transfer-Encoding: chunked
        Access-Control-Allow-Headers: X-Requested-With, Content-Type, Authorization, Date
        Access-Control-Allow-Methods: GET, HEAD, POST, PUT, DELETE, OPTIONS
        Access-Control-Allow-Origin: *
        Cache-Control: no-cache, no-store, must-revalidate
        Connection: keep-alive
        Content-Type: application/json
        Date: Fri, 20 May 2022 15:47:23 GMT
        Server: nginx/1.18.0
        
        278
        {"capabilities":{"m.room_versions":{"default":"9","available":{"1":"stable","2":"stable","3":"stable","4":"stable","5":"stable","6":"stable","org.matrix.msc2176":"unstable","7":"stable","8":"stable","9":"stable","org.matrix.msc2716v3":"unstable","org.matrix.msc3787":"unstable"},"org.matrix.msc3244.room_capabilities":{"knock":{"preferred":"7","support":["7","8","9","org.matrix.msc2716v3","org.matrix.msc3787"]},"restricted":{"preferred":"9","support":["8","9","org.matrix.msc3787"]}}},"m.change_password":{"enabled":true},"m.set_displayname":{"enabled":true},"m.set_avatar_url":{"enabled":true},"m.3pid_changes":{"enabled":true}}}
        0
        
    server.go:163: Creating room !0:host.docker.internal:42517 with version 9
    federation_room_join_partial_state_test.go:65: Making POST request to http://localhost:49286/_matrix/client/r0/join/%210:host.docker.internal:42517?server_name=host.docker.internal%3A42517 (syt_YWxpY2U_oGHIbbKayEgVymeIvaNI_0QUzop)
2022/05/20 16:47:23 Not a partial-state request: got , want true
2022/05/20 16:47:24 Not a partial-state request: got , want true
2022/05/20 16:47:25 Not a partial-state request: got , want true
2022/05/20 16:47:27 Not a partial-state request: got , want true
    client.go:544: POST hs1/_matrix/client/r0/join/!0:host.docker.internal:42517 => 502 Bad Gateway (3.965565545s)
    federation_room_join_partial_state_test.go:65: HTTP/1.1 502 Bad Gateway
        Transfer-Encoding: chunked
        Access-Control-Allow-Headers: X-Requested-With, Content-Type, Authorization, Date
        Access-Control-Allow-Methods: GET, HEAD, POST, PUT, DELETE, OPTIONS
        Access-Control-Allow-Origin: *
        Cache-Control: no-cache, no-store, must-revalidate
        Connection: keep-alive
        Content-Type: application/json
        Date: Fri, 20 May 2022 15:47:27 GMT
        Server: nginx/1.18.0
        
        44
        {"errcode":"M_UNKNOWN","error":"Failed to send_join via any server"}
        0
        
    federation_room_join_partial_state_test.go:65: CSAPI.MustDoFunc response return non-2xx code: 502 Bad Gateway - body: {"errcode":"M_UNKNOWN","error":"Failed to send_join via any server"}

obtained with

COMPLEMENT_DEBUG=1 WORKERS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh -run TestSyncBlocksDuringPartialStateJoin

[squahtx]

At the bottom of scripts-dev/complement.sh, we have:

go test -v -tags synapse_blacklist,msc2716,msc3030,faster_joins -count=1 "${extra_test_args[@]}" "$@" ./tests/...

If we omit the faster_joins tag when WORKERS=1, we should be able to disable the failing tests until fast room joins is worker-ready.

[reivilibre]

Aha, thank you very much for spotting that @squahtx!