On 06/06/2012 02:20 PM, John Eckersberg wrote:
On Wed, 06 Jun 2012 16:20:17 -0400, John
Eckersberg<jeckersb(a)redhat.com> wrote:
> On Wed, 6 Jun 2012 12:51:25 -0700, Richard Su<rwsu(a)redhat.com> wrote:
>> We support systems that are not systemd aware. Resurrected the old
>> SysV init script to support those systems.
>>
>> Modified conductor spec file to install either the systemd or SysV
>> init configuration files depending on what the os supports.
>> ---
>> aeolus-conductor.spec.in | 31 +++++++++++++++++-
>> conf/conductor-delayed_job | 77 ++++++++++++++++++++++++++++++++++++++++++++
>> 2 files changed, 107 insertions(+), 1 deletions(-)
>> create mode 100755 conf/conductor-delayed_job
>>
>> diff --git a/aeolus-conductor.spec.in b/aeolus-conductor.spec.in
>> index 4c5aad5..9bee661 100644
>> --- a/aeolus-conductor.spec.in
>> +++ b/aeolus-conductor.spec.in
>> @@ -9,6 +9,12 @@
>> %global rubyabi 1.8
>> %endif
>>
>> +%if %((test -d /etc/systemd&& echo 1) || echo 0)
>> +%global systemd 1
>> +%else
>> +%global systemd 0
>> +%endif
>> +
> The alarms in my head say this might not work depending on what context
> you're building in. I'm not so sure you're going to have /etc/systemd
> present in the buildroot for all systemd-enabled targets under mock (and
> therefor in koji and brew). You may be better off conditionalizing this
> using %{?fedora} and %{?rhel} and checking versions that way.
>
> /me goes off to look into that...
Verified, at least in a clean mock root for fedora-16-x86_64:
[jeckersb@baozi ~]$ test -d /var/lib/mock/fedora-16-x86_64/root/etc/systemd
[jeckersb@baozi ~]$ echo $?
1
So you won't use the systemd service unit on f16, even though it is
supported.
Good catch. I will send a updated patch to use version numbers.