Failed RPC with distressed network causes application to stop functioning

Description of the bug / feature

Flow application stops working when network exhibits high jitter & latency.

Logs include multiple instances of Unexpected message id from the client. Expected sync id: X, got Y, often in bursts over the course of a few seconds.

When digging into this, I came across this comment: https://github.com/vaadin/flow/blob/a45fe2ee201fe0587e381a11a6d38744403282c3/flow-server/src/main/java/com/vaadin/flow/server/communication/ServerRpcHandler.java#L293-L301 I thought this bug report may provide the desired data. A network diagnosis at the user’s site showed an average latency of 108ms and an average jitter of 157ms (very high). Network requests were taking as long as 2200 ttfb. Average bandwidth was about 10Mbps

Minimal reproducible example

Flow application - Push may be required.

Expected behavior

I’m not sure. Perhaps it would be useful to track over time - if it’s intermittent, then resync. If a client is experiencing this frequently, then a notification of an unreliable network - like one might see in a video call.

Actual behavior

User is notified of an internal error. The application stops functioning.

Versions:

- Vaadin / Flow version: 14.4.0
- Java version: openjdk 11.0.9.1
- OS version: AWS Linux
- Browser version (if applicable): Chrome on Windows 87.0.4280.141
- Application Server (if applicable): Spring Boot
- IDE (if applicable): N/A

[epricer]

I will try to post something within a week.

[miguelatvaadin]

I will close this issue as no answer since March, 25. Please feel free to post a comment or reopen when there is new info. Thanks