Project

General

Profile

Actions

action #59852

closed

auto_review:"Failed to find an available port: Address already in use" Fix issues in Mojo::IOLoop::ReadWriteProcess leading to broken workers due to leftover QEMU processes

Added by mkittler over 4 years ago. Updated almost 4 years ago.

Status:
Resolved
Priority:
Normal
Assignee:
Category:
Regressions/Crashes
Target version:
-
Start date:
2019-11-14
Due date:
% Done:

0%

Estimated time:

Description

The worker's logic to prevent job execution when the QEMU from the previous test run hasn't been terminated was broken (see #56993).

This has already been fixed but showed that Mojo::IOLoop::ReadWriteProcess is unable to stop QEMU in all cases. Even with version 0.24 I could reproduce that very sporadic issue (on my Tumbleweed system).

I also observed that the CGroup feature might be broken, see my comment in #56993:

With this fix it at least creates the cgroup and passes it to the Mojo::IOLoop::ReadWriteProcess::Container thing. But I've just checked whether this has any actual influence on the cgroup tree (systemd-cgls) but it seems isotovideo and its forks including QEMU are just within the usual session-2.scope.

The could be the cause why Mojo::IOLoop::ReadWriteProcess is sometimes unable to stop QEMU.

Actions

Also available in: Atom PDF