Re: [PATCH] http admin commands

Am 07.07.2009 um 18:10 schrieb Alejandro Guerrieri:

> Alex,
>
> I agree on having full support for reload on all kannel's  
> configuration options, of course.
>
> However, that could pose a problem if you play "too hard" with the  
> config files. Some sort of checksum on groups would be required to  
> detect which things changed and also which things were removed  
> (thinking aloud here, I didn't really analyzed it in depth). Also, a  
> full syntax verification would be needed before applying the changes  
> (nagios-style).
>
> There are some situations where a simple add-smsc or remove-smsc  
> could come in handy: I could need to add just a new smsc connection  
> for testing purposes and remove it soon afterwards (or swapping it  
> for another one known to work). I could have made changes on many  
> connections, but I'd need to test it one by one... etc.
>
> Even having full-reload capabilities, I think add-smsc and remove- 
> smsc are useful addition to the command list.
>
> Right now, if I'm not wrong only the white/black list can be  
> reloaded on the fly. I'll analyze the rest of the code and see how  
> to implement full-reload capabilities (after providing the userguide  
> patch for the current patch).

white/black list reload is currently buggy (race condition). Stipe  
should fix it but still didn't fix it... Stipe?

>
> Regards,
> --
> Alejandro Guerrieri
> [email protected]
>
>
>
> On 07/07/2009, at 17:45, Alexander Malysh wrote:
>
>> yes, I see your concern :)
>>
>> But I think that if we want reload it should work for all bearerbox  
>> parts and not only smsc, something like config diff and then reload
>> changed parts...
>>
>> Alex, don't get me wrong, I'm not against your patch, I'm just  
>> don't like partial support :)
>> Please provide userguide patch and it could be commited...
>>
>> Thanks,
>> Alex
>>
>> Am 07.07.2009 um 15:36 schrieb Alejandro Guerrieri:
>>
>>> Example #1: You have a huge queue (hundreds of thousands) of  
>>> outgoing messages going out. Restarting the whole service can take  
>>> tens of minutes, specially when you're repeatedly tweaking those  
>>> services to try to send as fast as possible. This happened to me  
>>> before: the carrier was dropping the link when we were sending too  
>>> many messages at once, so we had to trial and error with max- 
>>> pending-submits and throughput, and each restart took about 15  
>>> minutes, so you get the idea...
>>>
>>> Example #2: You have 20+ smsc binds. Turning all those down and up  
>>> again repeatedly can annoy a carrier or two, and also hurt your  
>>> service level agreements if any.
>>>
>>> I can foresee another scenarios where this would be valuable.
>>>
>>> Regards,
>>> --
>>> Alejandro Guerrieri
>>> [email protected]
>>>
>>>
>>>
>>> On 07/07/2009, at 15:13, Alexander Malysh wrote:
>>>
>>>> Hi Alex,
>>>>
>>>> patch looks ok but I don't really see why you should be unable to  
>>>> restart the whole bearerbox...
>>>> and I think reload should work for all bearerbox parts...
>>>>
>>>> therefore I'm +0 for this patch.
>>>>
>>>> Thanks,
>>>> Alex
>>>>
>>>> Am 07.07.2009 um 08:14 schrieb Alejandro Guerrieri:
>>>>
>>>>> This patch improves the http-based admin commands and fixes some  
>>>>> quirks:
>>>>>
>>>>> 1. It adds 2 new commands: add-smsc and remove-smsc that allows  
>>>>> for dynamic adding/removing of smsc's without restarting kannel  
>>>>> nor any other smsc bind.
>>>>>
>>>>> 2. start-smsc now reloads the smsc list from the config file, so  
>>>>> you can stop-smsc and then start-smsc to reload the  
>>>>> configuration. The only caveat is: do not change the id or admin- 
>>>>> id (in that case you'll have to remove-smsc with the old id/ 
>>>>> admin-id and add-smsc with the new one instead).
>>>>>
>>>>> 3. Fixes a bug on stop-smsc and start-smsc. Before, you could  
>>>>> start/stop non-existant smsc's, and despite the request would be  
>>>>> ignored, the interface would say that the smsc was stopped/ 
>>>>> started. Now it will return an error and log the reason.
>>>>>
>>>>> 4. Fixes a bug on the id/admin-id functionality when stopping/ 
>>>>> starting a group with 2 or more entries with the same id and  
>>>>> different admin-id's. Before, the list would get corrupted when  
>>>>> trying to start again, now it works as expected.
>>>>>
>>>>> Userguide patch coming soon :)
>>>>>
>>>>> Please review in the meantime.
>>>>>
>>>>> Regards,
>>>>> --
>>>>> Alejandro Guerrieri
>>>>> [email protected]
>>>>>
>>>>>
>>>>> <kannel-http-admin.patch.zip>
>>>>
>>>
>>
>