Re: generation of virtproxd socket files

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mon, Feb 08, 2021 at 04:09:16PM -0700, Jim Fehlig wrote:
Hi All,

I received a report [1] and verified that virtproxyd*.socket files have broken
syntax. E.g. from virtproxyd.socket

[Unit]
Description=Libvirt proxy local socket
Before=virtproxyd.service
libvirtd.socket libvirtd-ro.socket libvirtd-admin.socket libvirtd-tcp.socket
libvirtd-tls.socket

virtproxyd.socket should 'Conflicts' with the libvirtd sockets. I suspect this
regressed in the switch to meson. I checked a libvirt 6.0.0 installation and
indeed it has

Conflicts=libvirtd.socket libvirtd-ro.socket libvirtd-admin.socket
libvirtd-tcp.socket libvirtd-tls.socket


It looks like the issue might have various root causes as I am unsure myself
about what exactly was the idea behind some of the actions in the code.  But all
the .socket files are generated from `src/remote/libvirtd.socket.in` and the line which causes trouble for you is probably this one:

  https://gitlab.com/libvirt/libvirt/-/blob/a619e28dba8c/src/remote/libvirtd.socket.in#L4

This is used in `src/meson.build` to generate socket files for all the services:

  https://gitlab.com/libvirt/libvirt/-/blob/a619e28dba8c/src/meson.build#L792-833

The proxy daemon service lists the other sockets as dependencies here (even though it uses variable clearly named `*_conflicts`):

  https://gitlab.com/libvirt/libvirt/-/blob/a619e28dba8c/src/remote/meson.build#L233

and I'm not sure whether that should be prefixed with "Conflicts=" or whether
the `conflicts` field should come into play, although it looks like it is not
used anywhere else:

  https://gitlab.com/libvirt/libvirt/-/blob/a619e28dba8c/src/meson.build#L204

So long story short: I do not have a right solution for you, but I hope this can
help you figure out where to go from here.

Actually, let me check if the `deps` are used anywhere else for a unit... yep,
it is, for virtxend and it actually uses it for the whole line it needs in that
context.  So maybe a simple patch like this could help?

diff --git i/src/remote/meson.build w/src/remote/meson.build
index 9ad2f6ab1c26..0a188268b58b 100644
--- i/src/remote/meson.build
+++ w/src/remote/meson.build
@@ -230,7 +230,7 @@ if conf.has('WITH_REMOTE')
       'name': 'Libvirt proxy',
       'sockprefix': 'libvirt',
       'sockets': [ 'main', 'ro', 'admin', 'tcp', 'tls' ],
-      'deps': libvirtd_socket_conflicts,
+      'deps': 'Conflicts=' + libvirtd_socket_conflicts,
     }

     openrc_init_files += {
--

I'm not well versed in the systemd unit files, so feel free to adjust for
whatever feels right.

I hope that helps, have a nice day.
Martin

I blame it on Monday and my mind stuck in the weekend, but I spent too much time
trying to figure out how those socket files are generated before writing this
mail. It would be much appreciated if someone can give me a nudge in the right
direction :-).

Regards,
Jim

[1] https://bugzilla.opensuse.org/show_bug.cgi?id=1181838

Attachment: signature.asc
Description: PGP signature


[Index of Archives]     [Virt Tools]     [Libvirt Users]     [Lib OS Info]     [Fedora Users]     [Fedora Desktop]     [Fedora SELinux]     [Big List of Linux Books]     [Yosemite News]     [KDE Users]     [Fedora Tools]

  Powered by Linux