Re: Implementing a set-before-get-behaviour with presets
On Mon, 2016-10-10 at 16:12 +1100, scl...@tpg.com.au wrote: > Can you explain this "takes priority in case of a configuration > conflict", perhaps with an example? Say you have two presets that set the same parameter. The value from the preset with the highest weight will be the one pushed to the device. > Also, If we want to have an ordering of presets, (some preset to be > be tested before others, and knowing that caching is in effect, and > settings not being refreshed) would we have to delete them an add > them back in a certain order to obtain this ordering effect? Not sure I understand what you're trying to do, but the ordering of presets shouldn't matter. All preset preconditions are tested atomically and the ones that match will then be applied. -- Zaid Abdulla z...@genieacs.com ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
Hi Zaid, In regards to this answer on ordering: On 22/03/16 03:47, Zaid Abdulla wrote: On Fri, 2016-03-18 at 14:40 +0100, Oliver Kraitschy wrote: - How is weight evaluated and used? Does it determine an order of execution or a precedence of execution? It doesn't determine the ordering, it determines which preset takes proirity in case of configuration conflicts. Can you explain this "takes priority in case of a configuration conflict", perhaps with an example? Also, If we want to have an ordering of presets, (some preset to be be tested before others, and knowing that caching is in effect, and settings not being refreshed) would we have to delete them an add them back in a certain order to obtain this ordering effect? Cheers, Stuart ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
On Thu, May 12, 2016 at 04:06:24AM +0300, Zaid Abdulla wrote: > > It shouldn't take that long! Try setting the config option > "GET_PARAMETER_NAMES_DEPTH_THRESHOLD" to 2 or 3 and see if that helps. So how long should it take? ;-) I'm already using "GET_PARAMETER_NAMES_DEPTH_THRESHOLD" : 3, "SESSION_TIMEOUT": 90 to prevent timeouts. But i suppose that we can optimize our cwmp scripts for performance. Which TR069 client do you use? > Yes, I believe you will be able to achieve something like that in v1.1. Looking forward to that! Greetings, Oliver ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
Yeah. I had to set it to 3 otherwise I'd get timeouts. -dan > On May 11, 2016, at 6:46 PM, Zaid Abdulla wrote: > >> On Wed, 2016-05-11 at 18:35 -0800, Dan Morphis wrote: >> One of our CPEs takes that long. It has a slow processor and large >> object model. > > I'm guessing you've already tried "GET_PARAMETER_NAMES_DEPTH_THRESHOLD" > and it didn't help? I've had a similar issue with one of those devices > with limited resources. IIRC, that option helped as it reduced the > response for GetParameterNames to smaller chunks. > > -- > Zaid Abdulla > z...@genieacs.com > > ___ > Users mailing list > Users@lists.genieacs.com > http://lists.genieacs.com/mailman/listinfo/users ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
On Wed, 2016-05-11 at 18:35 -0800, Dan Morphis wrote: > One of our CPEs takes that long. It has a slow processor and large > object model. I'm guessing you've already tried "GET_PARAMETER_NAMES_DEPTH_THRESHOLD" and it didn't help? I've had a similar issue with one of those devices with limited resources. IIRC, that option helped as it reduced the response for GetParameterNames to smaller chunks. -- Zaid Abdulla z...@genieacs.com ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
-dan > On May 11, 2016, at 5:06 PM, Zaid Abdulla wrote: > >> On Mon, 2016-05-09 at 15:37 +0200, Oliver Kraitschy wrote: >> The initial refreshObject is working fine. The problem is that it >> takes ~ 10 min to fetch all parameters (~1500). We want to set >> specific parameters first and fetch all parameters afterwards as set- >> before-get implies. > > It shouldn't take that long! Try setting the config option > "GET_PARAMETER_NAMES_DEPTH_THRESHOLD" to 2 or 3 and see if that helps. One of our CPEs takes that long. It has a slow processor and large object model. ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
On Mon, 2016-05-09 at 15:37 +0200, Oliver Kraitschy wrote: > The initial refreshObject is working fine. The problem is that it > takes ~ 10 min to fetch all parameters (~1500). We want to set > specific parameters first and fetch all parameters afterwards as set- > before-get implies. It shouldn't take that long! Try setting the config option "GET_PARAMETER_NAMES_DEPTH_THRESHOLD" to 2 or 3 and see if that helps. > Additionally we want to track and store the parameter settings of a > specific device in the ACS in a kind of device profile so that the > device settings can be easily restored if a device gets replaced > (because of hw failureetc). At the moment i misuse presets as device > profiles. But i had to implement a mechanism to keep them up to date > if a parameter is modified through the webinterface of the device (a > notification is sent by easycwmp and the preset is updated > accordingly). > > How would you implement that? Will there be any functionality in the > upcoming release to implement such a behaviour (and set-before-get)? Yes, I believe you will be able to achieve something like that in v1.1. -- Zaid Abdulla z...@genieacs.com ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
On Fri, Apr 22, 2016 at 03:03:10AM +0300, Zaid Abdulla wrote: > On Fri, 2016-04-15 at 15:54 +0200, Oliver Kraitschy wrote: > > i have the problem that i can't set a parameter (with a preset) which > > is unknown to the ACS (this happens before any refreshObject has > > taken place, thus set-before-get) > > The ACS should fetch the entire data model of the device when it > connects for the first time. Is the initial refreshObject failing for > some reason? Hello Zaid, sorry for the delay. The initial refreshObject is working fine. The problem is that it takes ~ 10 min to fetch all parameters (~1500). We want to set specific parameters first and fetch all parameters afterwards as set-before-get implies. I solved the problem now by creating a refresh task for all parameters included in that preset so i can set them with the preset first and fetch all parameters afterwards. Additionally we want to track and store the parameter settings of a specific device in the ACS in a kind of device profile so that the device settings can be easily restored if a device gets replaced (because of hw failure etc). At the moment i misuse presets as device profiles. But i had to implement a mechanism to keep them up to date if a parameter is modified through the webinterface of the device (a notification is sent by easycwmp and the preset is updated accordingly). How would you implement that? Will there be any functionality in the upcoming release to implement such a behaviour (and set-before-get)? Greetings, Oliver ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
On Fri, 2016-04-15 at 15:54 +0200, Oliver Kraitschy wrote: > i have the problem that i can't set a parameter (with a preset) which > is unknown to the ACS (this happens before any refreshObject has > taken place, thus set-before-get) The ACS should fetch the entire data model of the device when it connects for the first time. Is the initial refreshObject failing for some reason? Zaid ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
Hello, when implementing the set-before-get-behaviour, i have the problem that i can't set a parameter (with a preset) which is unknown to the ACS (this happens before any refreshObject has taken place, thus set-before-get). The problem could be that a presets discrepancy can't be identified without knowing the previous value. Additionally it seems i can't refresh only the sub-nodes where i want to set my parameters. How can i implement this set-before-get-behaviour? Any hints are greatly appreciated. Greetings, Oliver ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
On Tue, 2016-03-22 at 13:40 +0100, Oliver Kraitschy wrote: > Should i set the interval of the refresh configuration in the first > preset to a higher value, 5 seconds for example? Yes, sorry I forgot to mention that. It's a hacky workaround but it should do the job. Zaid ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
On Fri, 2016-03-18 at 14:40 +0100, Oliver Kraitschy wrote: > Why doesn't get the second preset executed? That's due to caching. It's a limitation in this version and is unfortunately difficult to fix without introducing undesired side effects. A quick workaround is to add a "refersh" configuration in your first preset with interval of 1 second. This will force the cache duration to not exceed 1 second which should hopefully be enough time for it to expire within the same session. > - How are tags supposed to be used in preconditions? Can i use > several tags in one precondition? Like: Tags = > "configured,refreshed"? Do i need a separate precondition for each > tag? Add multiple tag filters in your precondition, one for each tag. > - How is weight evaluated and used? Does it determine an order of > execution or a precedence of execution? It doesn't determine the ordering, it determines which preset takes proirity in case of configuration conflicts. > - If i add a refresh configuration with an interval (age) of 0 > seconds, does that mean it's executed only once? > - If i don't enter an interval for a refresh configuration, the cwmp > worker dies. It means it will be executed infinite number of times :) Value should be 1+. > When is a preset executed? After the inform of a device? Right after inform only when there are no pending tasks in queue. Zaid ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
Re: Implementing a set-before-get-behaviour with presets
On Fri, Mar 18, 2016 at 02:40:30PM +0100, Oliver Kraitschy wrote: > Hello, > > i'm trying to achieve a get-before-set behaviour with genieacs v1.0.6. I meant to write "set-before-get behaviour", of course. Greetings, Oliver ___ Users mailing list Users@lists.genieacs.com http://lists.genieacs.com/mailman/listinfo/users
