[Freeipa-devel] [PATCH 0069] ipa-nis-manage enable: change service name from 'portmap' to 'rpcbind'

[Martin Basti]

On 10.05.2016 15:36, Abhijeet Kasurde wrote:
>
>
> On 05/10/2016 06:48 PM, Martin Basti wrote:
>>
>>
>>
>> On 10.05.2016 14:50, Gabe Alford wrote:
>>> On Tue, May 10, 2016 at 6:47 AM, Martin Basti <mbasti at redhat.com> wrote:
>>>
>>>
>>>
>>>     On 10.05.2016 14:42, Gabe Alford wrote:
>>>>     On Tue, May 10, 2016 at 6:26 AM, Martin Basti
>>>>     <mbasti at redhat.com> wrote:
>>>>
>>>>
>>>>
>>>>         On 10.05.2016 14:13, Gabe Alford wrote:
>>>>>         On Tue, May 10, 2016 at 2:00 AM, Martin Basti
>>>>>         <mbasti at redhat.com> wrote:
>>>>>
>>>>>
>>>>>
>>>>>             On 04.05.2016 15:14, Gabe Alford wrote:
>>>>>>             On Tue, May 3, 2016 at 11:17 PM, Abhijeet Kasurde
>>>>>>             <akasurde at redhat.com> wrote:
>>>>>>
>>>>>>                 Hi Gabe,
>>>>>>
>>>>>>                 I am wondering, how are we handling
>>>>>>                 "CalledProcessError" exception ?
>>>>>>
>>>>>>
>>>>>>             I am not sure 100% what you are asking, but from what
>>>>>>             I understand, the "CalledProcessError" exception is
>>>>>>             when a process returns a non-zero exit status.
>>>>>>             However when running 'ipa-nis-manage enable', an
>>>>>>             exception is never hit even if portmap is not
>>>>>>             installed, hence portmap always being enabled.
>>>>>>
>>>>>>             So it seems that if the process is not installed,
>>>>>>             "CalledProcessError" doesn't catch an error.
>>>>>>
>>>>>>             Gabe
>>>>>             Hello,
>>>>>
>>>>>             portmap.enable() may raise the "CalledProcessError" in
>>>>>             case that systemct enable failed and we should catch
>>>>>             this exception and handle it in the same way as it is
>>>>>             done now. i.e catch that exception and set proper
>>>>>             return state.
>>>>>
>>>>>             Martin^2
>>>>>
>>>>>
>>>>>         Shouldn't "CalledProcessError" raise an exception in this
>>>>>         case? In my testing, it doesn't seem to raise an exception
>>>>>         when the service does not even exist on the system.
>>>>>
>>>>>         Gabe
>>>>>
>>>>         You are right, there is try-except-pass, so no exception
>>>>         can be raised
>>>>
>>>>         def __enable(self, instance_name=""):
>>>>              try:
>>>>                  ipautil.run([paths.SYSTEMCTL,"enable",
>>>>                               self.service_instance(instance_name)])
>>>>              except ipautil.CalledProcessError:
>>>>                  pass
>>>>
>>>>
>>>>         Martin
>>>>
>>>>
>>>>     It is also the case for disable(), mask(), unmask(), etc.
>>>>     Should we update the exception in __enable() or is there a
>>>>     reason that it just passes at exception?
>>>>
>>>>     Gabe
>>>
>>>     I dont think that we should chnge behavior there, what I'm
>>>     missing there is proper logging :) If you want you can create
>>>     ticket for it. Leave try-except-pass there, changing this may
>>>     affect a lot of places, and there is no time to fix it in 4.4
>>>     release.
>>>
>>>     Martin^2
>>>
>>>
>>> Sounds good. Do you also want to keep the try-except-pass in 
>>> ipa-nis-manage as well or does my patch suffice?
>>>
>>> Gabe
>>
>> I'm fine with your patch if Abhijeet agree, we can push it.
>> Martin^2
>>
> ACK.

Pushed to master: bede6c282e6d321c348dc2d33c6d1f9c14093a57
>>>
>>>>>
>>>>>>                 On 05/04/2016 09:17 AM, Gabe Alford wrote:
>>>>>>>                 Hello,
>>>>>>>
>>>>>>>                 Fix for https://fedorahosted.org/freeipa/ticket/5857
>>>>>>>
>>>>>>>                 Thanks,
>>>>>>>
>>>>>>>                 Gabe
>>>>>>>
>>>>>>>
>>>>>>                 Thanks,
>>>>>>                 Abhijeet Kasurde
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>
>>>
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/freeipa-devel/attachments/20160510/23d6c426/attachment.htm>