You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I observed that after creating a room (generating a lot of transaction traffic for each created event) that the events trickled through to the appservice rather than being sent in quick succession. Having looked at the logs, I could see timeouts for the /users/ endpoint.
While I believe the AS timing out the response to the /users/ endpoint is a bug on our side, I don't think Synapse should be queuing transactions behind it.
I think this might be the result of a recent change to authenticate more requests to the AS endpoints, but that is pure speculation.
Logs:
2022-10-12 09:49:16,326 - synapse.appservice.api - 129 - WARNING - notify_interested_services-3- query_user to http://appservice-service:9995/users/%40server%3Aserver.host received 504
2022-10-12 09:49:16,344 - synapse.http.client - 446 - INFO - as-sender-5- Received response to PUT http://appservice-service:9995/transactions/8?access_token=<redacted>: 200
....
2022-10-12 09:50:16,330 - synapse.http.client - 455 - INFO - notify_interested_services-3- Error sending request to GET http://appservice-service:9995/users/%40server%3Aserver.host?access_token=<redacted>: RequestTimedOutError 504: Timeout waiting for response from remote server
2022-10-12 09:50:16,330 - synapse.appservice.api - 129 - WARNING - notify_interested_services-3- query_user to http://appservice-service:9995/users/%40server%3Aserver.host received 504
2022-10-12 09:50:16,346 - synapse.http.client - 446 - INFO - as-sender-6- Received response to PUT http://appservice-service:9995/transactions/9?access_token=<redacted>: 200
The text was updated successfully, but these errors were encountered:
This issue has been migrated from #14152.
I observed that after creating a room (generating a lot of transaction traffic for each created event) that the events trickled through to the appservice rather than being sent in quick succession. Having looked at the logs, I could see timeouts for the /users/ endpoint.
While I believe the AS timing out the response to the /users/ endpoint is a bug on our side, I don't think Synapse should be queuing transactions behind it.
I think this might be the result of a recent change to authenticate more requests to the AS endpoints, but that is pure speculation.
Logs:
The text was updated successfully, but these errors were encountered: