Hi! Is there a possibility to influence the labels of a bug i.e. to ask the labeller to reconsider? There is this bug (Unexpected message id from the client (sync ids different) · Issue #12640 · vaadin/flow · GitHub) that is labelled Impact:Low Severity:Minor which is opened since 2+ years and (according to the comments) possibly impacting a lot of applications with high severity (error or empty page).
The forum isn’t the place for this. You should push up the issue - especially with findings on how to reproduce, test setup’s and / or even showing how this effects your users so that it can re-triaged.
So you mean more comments (from more users, including how to reproduce etc.) is the only way to get a re-triage?
That error is actually reported multiple times and has been discussed on many tickets. The error itself is not a bug as is, but logged as part of the failsafe mechanism in Vaadin.
When packets are sent back and forth between server and browser, they include serial number so that both parties can verify that no packet is being lost.
There are natural reasons for packet loss
- Wrong load balancer / proxy setup, e.g. without sticky sessions
- Firewall, VPN blocking packet due deep inspection, I have seen cases where packets have been blocked due JavaScript or JSON carried in them and too strict settings
- The above can also just delay packet arrival, especially virus sniffers are problematic
- Faulty router (yes, that has happened in real life, one customer had this and after router firmware update issue disappeared)
- Flake WiFi/Cellular connection (yes has happened)
- Bug in Vaadin (we have fixed many, thus most of the tickets discussing this are closed). Naturally this is the final option still which cannot be ruled out.
When communication goes un-sync Vaadin detects that and does resynchronize. That is done on purpose in order to restore the operation.
When you see this error, one should observe does it happen in correlation with some usage pattern of the application, frequency, when it is happening etc. All these can give some hints whether it is due some natural cause of the error.
As you see the list of the natural causes are all flaky, hence deterministic re-production may be difficult.
The cases where it has been a bug in Vaadin have been difficult to debug, but with rigorous attempt of following the statistics, close collaboration with the reporter, ruling out natural causes we have been able to find those and fixed them. Naturally we would like to hope that we have identified all of them, and only the natural causes remain.
1 Like
@sebastian.dietrich thanks for bringing it up, I’m going to re-triage and prioritise this issue. You said that empty page is shown for you?
If it’s more than a log message on server, that’s probably a bug in resynchronisation logic.
Could we have a chat and possible a video call with screen sharing, so that we can get more information?
1 Like
The bug was just an example - the main question was how to trigger re-evaluation of the labels. AFAICS this question is not yet answered.
Concerning the example bug: @Tatu2 thanks for that explanation. May I copy it to the bug (didn’t see any explanation for that behavior (at least not as clear as yours) in the bug itself)?
@mikhail.21 we get the bug logged irregularly on PROD (never on TEST). I personally have never experienced it. Users report that they get an empty page. I assume (reading @Tatu2 comment) that this only happens when the server restarted and the browser tab was not closed / user did not logout before (we stop the server on 30min. inactivity)
1 Like
Label re-evaluation and issue triage is a responsibility of Vaadin dev teams. We usually triage upcoming issue regularly. Old issues are re-evaluated occasionally based on direct request from customers or based on votes/comments in the GH ticket.
For this particular issue I’ve changed the labels to major severity and high impact and put it as a priority for upcoming dev iterations.
1 Like
What service on Google Cloud are you using?
Do you have high load?
1 Like
AppEngine and very low load - but that bug was just an example. If you have any ideas concerning that bug, please comment there
Thant’s strange. I run multiple applications on AppEngine.
Do you have session affinitiy set?
network:
session_affinity: true
We have
basic_scaling:
max_instances: 1
idle_timeout: 30m
So I am pretty sure that session_affinity would not help. Anyhow, it only occurs on PROD, while TEST has the same settings and runs as well on AppEngine.
Others mention in their bug as well, that more or less identical apps behave differently concerning this bug.