[pve-devel] [PATCH ha-manager] explicitly tell systemd debhelper to never stop watchdog-mux on upgrade

[Thomas Lamprecht]

On 1/25/18 3:01 PM, Thomas Lamprecht wrote:
> This allows to use an autogenerated postinst in a later patch
> 

Scratch that one, had an unclean working tree with an

Also=watchdog-mux.service

in the lrm/crm service Install section, which the dh_systemd_*
helper picked magically up. So the 'Also=' isn't needed and
without it my other series [1] is good without this, sorry for
the confusion :)


> Signed-off-by: Thomas Lamprecht <t.lamprecht at proxmox.com>
> ---
> 
> Needed for the postinst series I sent recently [1], else the removal
> of the trigger part there is not correct as it enables watchdog-mux
> restart again, which we really do not want.
> 
> [1]: https://pve.proxmox.com/pipermail/pve-devel/2018-January/030426.html
> 
>  debian/rules | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/debian/rules b/debian/rules
> index 468631f..0722c31 100755
> --- a/debian/rules
> +++ b/debian/rules
> @@ -20,6 +20,8 @@ override_dh_systemd_enable:
>  	dh_systemd_enable -ppve-ha-manager --name pve-ha-lrm pve-ha-lrm.service
>  
>  # restart LRM first, else single-node clusters with configured HA will deadlock
> +# ensure watchdog-mux gets started, but never stopped/restarted after upgrade
>  override_dh_systemd_start:
>  	dh_systemd_start -ppve-ha-manager --name pve-ha-lrm --restart-after-upgrade pve-ha-lrm.service
>  	dh_systemd_start -ppve-ha-manager --name pve-ha-crm --restart-after-upgrade pve-ha-crm.service
> +	dh_systemd_start -ppve-ha-manager --name watchdog-mux --no-stop-on-upgrade watchdog-mux.service
>