Hi,
On 26.10.2012 16:00, Pavel Březina wrote:
On 10/22/2012 03:46 PM, Jakub Hrozek wrote:
> On Mon, Oct 22, 2012 at 02:00:29PM +0200, Pavel Březina wrote:
>> On 10/19/2012 12:43 PM, Jakub Hrozek wrote:
>>> On Fri, Oct 19, 2012 at 10:19:25AM +0200, Pavel Březina wrote:
>>>> On 10/18/2012 03:24 PM, Simo Sorce wrote:
>>>>> On Thu, 2012-10-18 at 09:46 +0200, Pavel Březina wrote:
>>>>>>
https://fedorahosted.org/sssd/ticket/1437
>>>>>
>>>>> Fedora F15 has been EOLed and will not get updates, why bother with
>>>>> F15 ?
>>>>> Also F15 systemd was quite immature, and at the time guidelines
>>>>> mentioned you shouldn't change from sysv to systemd after GA.
>>>>> Did you mean to put F16 there ?
>>>>>
>>>>> Simo.
>>>>>
>>>>
>>>> OK. New patch is attached.
>>>
>>
>> Hi,
>>
>>>> %if (0%{?enable_experimental} == 1)
>>>> @@ -57,6 +61,12 @@ Requires(post): initscripts chkconfig
>>>> Requires(preun): initscripts chkconfig
>>>> Requires(postun): initscripts chkconfig
>>>>
>>>> +%if (0%{?use_systemd} == 1)
>>>> +Requires(post): systemd-units
>>>> +Requires(preun): systemd-units
>>>> +Requires(postun): systemd-units
>>>> +%endif
>>>
>>> I think here ^^ should be an %else that would otherwise Require
>>> initscripts. I don't think we need them in a pure systemd-enabled
>>> package.
>>
>> I'm no spec file expert, but everything is basically copy and paste
>> from our Fedora spec file so I assume it should be correct.
>>
>> The Fedora spec file says:
>> Requires(post): systemd-units initscripts chkconfig
>
> Right, I think the initscripts Requires is spurious,
I'm sending new patch that requires initscripts conditionally.
but we keep
> chkconfig to migrate from sysv scripts (which Pavel just showed me is
> forbidden by the Packaging Guidelines).
A few comments on the patch:
+Requires(post): chkconfig
+Requires(preun): chkconfig
+Requires(postun): chkconfig
Are you sure chkconfig is required in all of these? It is called only in
%triggerun, so Requires(post) should be enough AFAIK.
+%if (0%{?use_systemd} == 1)
+# Replace sysv init script with systemd unit file
+rm -f $RPM_BUILD_ROOT/%{_initrddir}/%{name}
+mkdir -p $RPM_BUILD_ROOT/%{_unitdir}/
+cp src/sysv/systemd/sssd.service $RPM_BUILD_ROOT/%{_unitdir}/
+%endif
I think this should be done by calling configure with the right options
instead:
%if (0%{?use_systemd} == 1)
%global with_initscript --with-initscript=systemd
--with-systemdunitdir=%{_unitdir}
%else
%global with_initscript --with-initscript=sysv
%endif
...
%configure \
...
%{with_initscript} \
...
+%preun
+if [ $1 = 0 ]; then
A nitpick: the test should read [ $1 -eq 0 ].
+%triggerun -- sssd < %{version}-%{release}
+if /sbin/chkconfig --level 3 sssd ; then
+ /bin/systemctl --no-reload enable sssd.service >/dev/null 2>&1 || :
+fi
+
+if /sbin/chkconfig --level 5 sssd ; then
+ /bin/systemctl --no-reload enable sssd.service >/dev/null 2>&1 || :
+fi
+
+/sbin/chkconfig --del sssd >/dev/null 2>&1 || :
I don't think this is right (both the sssd version and the body of the
scriptlet), read
<
http://fedoraproject.org/wiki/Packaging:ScriptletSnippets#Packages_migrat...;.
+%postun
+/bin/systemctl daemon-reload >/dev/null 2>&1 || :
+if [ $1 -ge 1 ] ; then
+ # On upgrade, reload init system configuration if we changed unit files
+ /bin/systemctl daemon-reload >/dev/null 2>&1 || :
+ /bin/systemctl try-restart sssd.service >/dev/null 2>&1 || :
+fi
There is no reason to call systemctl daemon-reload again inside the if.
Honza
--
Jan Cholasta