re-check/report old packagekit/systemd bugs

[github-actions[bot]]

Checked 182 URLs out of which 2 is/are invalid: https://bugs.freedesktop.org/show_bug.cgi?id=104552 : Internal Server Error https://bugs.freedesktop.org/show_bug.cgi?id=69575 : Internal Server Error

[martinpitt]

Yes, that server is no more. We can still see them in a web archive:

• http://web.archive.org/web/20220516113100/https://bugs.freedesktop.org/show_bug.cgi?id=104552 : This needs to be re-tested against a real current RHEL 9 installation with pending errata, and if necessary, re-reported.

• http://web.archive.org/web/20220523220015/https://bugs.freedesktop.org/show_bug.cgi?id=69575 : This bug is well described both in the bug report and also in our sources, even with an alternative piece of code. This can be tested with window.debugging = "dbus" and checking the signals.

[martinpitt]

The server is back. However, let's still re-test these old bugs, and re-report them against current systemd if necessary -- or give up and just declare that the behaviour.

[jelly]

One re-checked and closed https://github.com/cockpit-project/cockpit/pull/20957

[jelly]

I re-renabled our commented out code:

    // This is what we want to do:
    // systemd_manager.addEventListener("UnitNew", function (event, unit_id, path) {
    //     if (unit_id == name)
    //         refresh();
    // });

And quickly CPU usage goes up and busctl monitor logs tons of dbus requests.

[martinpitt]

Thanks @jelly ! So let's re-report this to the current systemd bug tracker and update the link.

[jelly]

Thanks @jelly ! So let's re-report this to the current systemd bug tracker and update the link.

There is no need for that, this is a known (unfixable?) race in systemd. The solution would be to ignore updates while we get property updates during a refresh().

https://github.com/systemd/systemd/issues/570#issuecomment-125334529