Posts: 12
Threads: 4
Joined: Aug 2017
Reputation:
0
Hi,
Does LM5 have a parameter for setting the maximum polling rate for KNX objects via multicast?
I've noticed that LM5 really pushes the capacity of the sublines.
Posts: 4572
Threads: 23
Joined: Aug 2017
Reputation:
206
(18.12.2018, 09:15)gs@el-24.no Wrote: Hi,
Does LM5 have a parameter for setting the maximum polling rate for KNX objects via multicast?
I've noticed that LM5 really pushes the capacity of the sublines.
Hi
LM does not pool any object unless you program it like that. It monitors live every telegram.
BR
------------------------------
Ctrl+F5
Posts: 4572
Threads: 23
Joined: Aug 2017
Reputation:
206
How do you do it?
------------------------------
Ctrl+F5
Posts: 4572
Threads: 23
Joined: Aug 2017
Reputation:
206
Then do it via script.
Tag all objects which you want to pool with tag 'Read' and make a schedule script which will run as often as you want. just make sure it will not run faster then time of execution of the script. 1000 X Delay
Code:
myobjects = grp.tag('Read')
for _, object in pairs(myobjects) do
-- send read request to object
grp.read(object.address)
os.sleep(1) --dealy in seconds
end
You can add also to start up script to read on boot.
------------------------------
Ctrl+F5
Posts: 4572
Threads: 23
Joined: Aug 2017
Reputation:
206
In here os.sleep(1) --dealy in seconds you specify how often you will make read request and it will go over all objects one by one with this delay.
Obviously disable all the previous pool settings.
I deleted not needed log in original script
------------------------------
Ctrl+F5
Posts: 7720
Threads: 42
Joined: Jun 2015
Reputation:
446
Keep in mind that with 1000 objects and 1 second delay each object will be polled once in 17 minutes. The question is why do you need polling instead of devices sending object values on change and/or by their own timer?
Posts: 7720
Threads: 42
Joined: Jun 2015
Reputation:
446
Ok then use what Daniel suggested. You can lower delay in os.sleep to 0.5 or less, depending on overall bus/system load.