After setting the configuration of slow poll within the element, I can still see in the streamviewer that groups are being polled normally.
When double checking the Protocol Pendingcalls (available in the SLNetClient Test Tool), I found that the slow poll feature will not start until the stack is empty.
However, for larger protocols it doesn't seem to be reaching the slow polling ever. After monitoring the Protocol Pendingcalls, I could see how groups and timers were in and out for a couple of times, making the stack to always be busy.
Is it just a matter of waiting until everything in the stack goes away? Or is there a misconfiguration with the element/protocol?
The protocol used in this test was CEFD H SNMP 1.5.1.8
More information about the element configuration can be found in the DataMiner Help: slow poll.
Looking at your element configuration, your element should go into slow poll mode after 30s of timeout. Does your element actually go into timeout or are only some requests timing out?
Once slow poll is actived, the ping group is executed until the device responds again. This needs to be configured in the protocol. More info can be found in the DataMiner Development Library: ping groups and ping group. As soon as the device responds ok to this request, the element will start polling again as well.
Could it be in your case that the ping group is not correct in the protocol? Another option could be that the device immediately responds to your ping group and the element starts polling regularly immediately?
Yes the element goes fully into timeout as the device is disconnected.
You are correct about the ping group: it is not properly configured. As the DataMiner Development Library states, the first group must be a polling group and not poll action group (which is the case for this driver)