On 09/21/2011 01:17 PM, Daniel P. Berrange wrote:
On Wed, Sep 21, 2011 at 12:46:33PM +0200, Peter Krempa wrote:
> On some systems init scripts are installed along with upstart . This may
> cause trouble if user tries to restart/stop a instance of libvirtd
> managed with upstart with init script.
>
> This patch adds check for a started libvirtd managed by upstart and
> fails the init script.
>
> [root@localhost ~]# initctl status libvirtd
> libvirtd start/running, process 3001
> [root@localhost ~]# service libvirtd restart
> Stopping libvirtd daemon: error: libvirtd is managed by upstart and
> started, use initctl instead
>
> If libvirtd is not managed by upstart or is stopped, init script works
> normaly and allows the user to manage the service.
>
>
https://bugzilla.redhat.com/show_bug.cgi?id=728153
> ---
> daemon/libvirtd.init.in | 18 ++++++++++++++++++
> 1 files changed, 18 insertions(+), 0 deletions(-)
NACK
- Hosts should not have both the SysV initscript& upstart script
installed at the same time
Well, they should not, but they do. And vdsm causes
that upstart
is used for libvirt even if rest of the system uses SysV.
Vdsm requires that libvirtd is restarted in case of a crash/quit and
abuses upstart and his "respawn" stanza.
- Even if both are installed together, there is no reason why a
libvirtd started by upstart cannot be stopped be SysV init and
vica-verca, since libvirtd puts its pidfile in the same place
regardless of how it is launched.
Because the upstart script states that the
libvirtd process should respawn
as soon as it exits, it's not possible for a SysV script to terminate
libvirtd.
Stop operation of the SysV script kills the libvirtd process and soon
after that
it's restarted again by upstart as configured.
Another way aroun this would be to detect if libvirtd is managable via
upstart at the beginning of the SysV script and hand control over to
upstart if this is the case.
Peter
Daniel