-
Notifications
You must be signed in to change notification settings - Fork 2.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
02-subpub-qos1-message-expiry.py
test regression
#3124
Comments
I'm running into something possibly related, and wanted to quickly point out it appears to happen in your screenshot above. If I need to reproduce it in something like Python, I can do that etc. When you're publishing from "helper" while "subpub-qos0-test" is subscribed, you're getting an "rc16" on the PUBACK, which I believe is a 0x10 from the MQTT specs for "no matching subscribers" (despite the fact you can see someone is subscribed above etc.). I'm working on a project where we send sort of a "header" and then more chunks if someone is listening. When I seemingly moved to 2.0.19, this bit of code no longer worked and it never thinks anyone is subscribed. When I use the 2.0.18 docker image, I don't have this issue. |
Actually disregard. In your case it does appear it closed subpub-qos0-test closed its connection before the PUBLISH happened, so rc16 makes sense. Still not sure why I'm getting it in my code on 2.0.19 and not 2.0.18, but I will dig deeper. |
This is the same issue as your other report, so I'm going to close it. Thanks for the analysis! |
Not really. This report is "the test fails when it didn't in the last release". The other report is "some test programs use an undefined variable when they fail". They're related, but they're definitely different. |
I apologize if I added confusion to all of this. I did originally think this was the same thing I was seeing, but I did change my mind later in my 2nd comment. The logs show the subscriber disconnecting before the rc16 comes back (possibly disconnecting due to the Python exception and the missing proto_ver?). I originally missed this. If the subscriber is disconnected, I think rc16 is appropriate. To be clear, I can't say this is the same thing as my bug (like I originally thought), but I also obviously defer to your experience overall. I personally latched onto this one too quickly because I was Googling for "rc16" and saw a newly submitted bug for the new version 😄. |
On v2.0.19 the
02-subpub-qos1-message-expiry.py
test fails:(I've filed issue 3123 about the
proto_ver
issue, which is only apparent when the test fails)I bisected the failure to commit 1914b3e ("Use separate trees for shared/normal subs").
I've reconfirmed the bisect as well.
The bisect log is:
I build with
and then run the test suite with the same command with
test
appended.The text was updated successfully, but these errors were encountered: