Strange LwM2M registration response.

[FRDavid]

Original issue from the lightweight-iotagent project moved to this project

Hello,

Note: This issue only appears in the grunt tests carried out by lightweigh-iotagent, not in the lwm2m-node-lib. Since the problem must be solved in lwm2m-node-lib, the issue has been opened in the lwm2m-node-lib project.

We have detected that the process of register a device is not carried out correctly due to the appearance of strange messages. They could appear also in other cases also. These messages are interspersed in the communications and produces problems since they are using the MID corresponding to the real response.

Seems libcoap is not waiting long enough for lightweight-iotagent and lwm2m-node-lib to send the response and we didn't find a way to disable this auto-response from libcoap.

This image shows the strange message using the correct MID for an "Empty response" generated automatically by libcoap, then the correct response using a different MID, so the correct response can not be correctly processed by the device.

A way to solve this issue is introduce the parameter "piggybackReplyMs" in the coap server creation, this is done in the lwm2m-node-lib library.

serverInfo.server = libcoap.createServer({
type: config.serverProtocol,
proxy: true,
piggybackReplyMs: 500
});

The piggybackReplyMs default value for libcoap library is 50ms and it seems too short for the processing time carried out by the lightweight-iotagent/lwm2m-node-lib set so I increased this value to 500 to ensure at least basic cases, but this should be evaluated in a future in other conditions such as overloaded.

Next image shows the communication after introduce this option. Now is clean from strange messages.

Tests are passed correctly in both lwm2m-node-lib and lightweight-iot agent.

BR,
David

[FRDavid]

Since this issue was already solved and merged. I will close this issue.

[fgalan]

The fix is not mentioned in https://github.com/telefonicaid/lwm2m-node-lib/blob/develop/CHANGES_NEXT_RELEASE. Maybe it should be.

[FRDavid]

Should be, so @fgalan check the PR #119 recently added