aszlig

@e-alfred: Confirmed also in my automated tests, here's the log: https://headcounter.org/hydra/log/pzn7ja0w24dxx33cg2474z455pn5by4z-vm-test-run-nextcloud-upgrade.drv

[This commit](https://github.com/openlab-aux/vuizvui/commit/4ab1416c9ded3ef3bec0618e1f7dd838d09f97c7) should shed some light about this.

Some more thoughts on this: The naive way to solve this would be to just wrap `clone`/`clone3` (except if using `CLONE_FILES` or `CLONE_VM`), `daemon` and `fork`/`vfork` and make sure the...

@riedel: Just to make sure my assumptions are correct: If you uncomment the following line: https://github.com/nixcloud/ip2unix/blob/b54905a0699d2ff3f24d9e219655a3ad7c9b3d85/src/socket.cc#L352 ... does it work?

Comment moved to https://github.com/nixcloud/ip2unix/pull/23.

> **Please please consider adding a command-line switch to enable this behavior for users who need it!** Leaking socket inodes is better than not working at all. The issue is...

> That would be a very different piece of software than the current ip2unix. Not having an extra PID floating around is a big motivation to use ip2unix instead of...

> AF_INET will keep accepting connections when one of the two processes does a close() while the other is still in its accept() loop. You can't accept new connections on...

@riedel: I'd probably prefer a different key such as -r `abstract=name`, since prepending `\0` looks a little bit weird on the command line, but in general I'm :+1: on that...

> If I do that, it doesn't get unlinked, but then inbound connections hang forever instead of just failing to connect. Can you please supply a run of `ip2unix -vvvvv`...