Re: [PATCH] remote: systemd: Remove unix sockets from filesystem when disabling a '.socket' unit

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

 



On Tue, Jan 18, 2022 at 03:51:21PM +0100, Michal Prívozník wrote:
> On 1/18/22 15:38, Peter Krempa wrote:
> > The existence of the unix socket path is used by the remote driver to
> > determine whether modular daemons are in use, so if the socket file
> > stays behind and the user decided to switch from modular to monolithic
> > daemon which was socket activated, the remote driver will insist on
> > picking '/var/run/libvirt/virtqemud-sock', even when it's no longer in
> > use:
> > 
> >  # systemctl start libvirtd.service
> >  # virsh list
> >   Id   Name   State
> >  --------------------
> > 
> >  # systemctl stop libvirtd.service
> >  Warning: Stopping libvirtd.service, but it can still be activated by:
> >    libvirtd.socket
> >    libvirtd-ro.socket
> >    libvirtd-admin.socket
> >  # systemctl start virtqemud.socket
> >  # virsh list
> >   Id   Name   State
> >  --------------------
> > 
> >  # systemctl stop virtqemud.socket
> >  # systemctl start libvirtd.service
> >  # virsh list
> >  error: failed to connect to the hypervisor
> >  error: Failed to connect socket to '/var/run/libvirt/virtqemud-sock': Connection refused
> > 
> >  # virsh -c 'qemu:///system?socket=/var/run/libvirt/libvirt-sock' list
> >   Id   Name   State
> >  --------------------
> > 
> > Fix this by instructing systemd to delete the socket file when
> > deactivating the unit file for the socket.
> > 
> > Signed-off-by: Peter Krempa <pkrempa@xxxxxxxxxx>
> > ---
> >  src/remote/libvirtd.socket.in | 1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/src/remote/libvirtd.socket.in b/src/remote/libvirtd.socket.in
> > index 85b4aa800a..0f349656f5 100644
> > --- a/src/remote/libvirtd.socket.in
> > +++ b/src/remote/libvirtd.socket.in
> > @@ -9,6 +9,7 @@ Before=@service@.service
> >  ListenStream=@runstatedir@/libvirt/@sockprefix@-sock
> >  Service=@service@.service
> >  SocketMode=@mode@
> > +RemoveOnStop=yes
> 
> I beg your pardon? Systemd leaves a stale socket behind? Isn't this
> something that systemd needs to fix?

I was surprised but their justification is:

      "Normally, it should
       not be necessary to use this option, and is not recommended as
       services might continue to run after the socket unit has been
       terminated and it should still be possible to communicate with them
       via their file system node. Defaults to off."

I would have some agreement with that view if it were possible to
then start the .socket unit again, and still connect to the .service.
Starting the .socket, however, does an implicit delete.

So leaving the socket on disk is only useful if you want to stop the
.socket, but never start it again.

On balance I think libvirt's need to distinguish daemons is more
important, unless we change the probing logic to actually try a
connect() but I wasn't too enthusiastic about doing that.

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|




[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