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
Next steps for 4.2.1 #1998
Comments
Just to clarify, in the case of systemd socket activation, systemd initially creates the socket file, not Puma. |
Is that true before the first start of Puma? If so, maybe what we need to do is when Puma starts, check for whether the file(s) exist, and set a flag that is used to determine whether to unlink the file(s) on shutdown? Or, just not add them to the Binder @unix_paths array if they do exist? |
@MSP-Greg Correct. systemd socket unit, creates the socket(file or port). |
Sorry, I've got a really squirrely ruby master bug that I'm trying to understand, just happened with a build early this morning...
Sorry, being stupid. Deleting the UNIXSocket files in Puma is different than deleting the files in test teardown. Re JRuby, I've tried it with Actions (both 9.2.8.0 & head), and they freeze all the time, but sometimes have only a few failures/errors. Until they stop freezing, I don't think I can be of any help, but I'll certainly watch for the fix. |
I believe #1987 solves the original bug. #1992 needs to be modified to work with systemd, as it was done as a 'please try this' PR. Puma currently unlinks all binder UNIXSocket files on shutdown or (hard/exec) restart. Puma should only unlink UNIXSocket files that it creates, and leave the files created by systemd alone. We can change the Binder I think the changes will cause merge conflicts with #1987, or #1987 blocks #1992. |
Got it, definitely agree with this. |
#1988 (comment) moving disco with @MSP-Greg here to keep stuff on-topic on the bug report thread
This annoys me, but it might be true. It annoys me because if we create a file, we should clean it up when we don't need it anymore. I'll have to look at the original issue more to reflect. So, gut feeling, I don't like this course of action.
Dunno what you mean.
Yup. I think the socket-closed bug + #1986 is a good point release.
NO, NO GREEN CHECKS FOR YOU! </soup-nazi> Forgetting about JRuby dug us into a hole the last time we did this. For now, I want to feel the pain of JRuby constantly failing, so that when 4.2.1 is released JRuby is the first thing we want to go and fix.
If you really really want to we can allow_fail until 4.2.1 is released, but I'm going to add it back in as soon as that release is tagged.
The text was updated successfully, but these errors were encountered: