Hi
Client : Wakamma client
Server : Leshan Server
We have been observing that Leshan server stops sending packets during a firmware upgrade process.
It happens in situations as below -
Example sinario:
server sends packet #20
server does not receive an ACK for packet #20
server again sends the same packet: packet #20
server receives and ACK packet #20
server sends next packet : packet #21
server receives second ACK for second packet #20
server stops sending packets entirely.
Meanwhile client for some reason has not received packet #21.
Below from Coap messages log:
------------------------------------------------------------------------------------------------------------------------------------
Time Coap Mesg MID Token Options
------------------------------------------------------------------------------------------------------------------------------------
14:53:41.009 CON-PUT 10021 1ebdc8 Uri-Path: 5, 0, 6 - Block1: ? - Content-Format: 1544
14:53:41.199 ACK-2.04 10021 1ebdc8 Block1: ?
14:53:41.199 CON-PUT 10022 1ebdc8 Uri-Path: 5, 0, 6 - Block1: ? - Content-Format: 1544
14:53:41.388 ACK-2.04 10022 1ebdc8 Block1: ?
14:53:41.388 CON-PUT 10023 1ebdc8 Uri-Path: 5, 0, 6 - Block1: - Content-Format: 1544
14:53:43.577 CON-PUT 10023 1ebdc8 Uri-Path: 5, 0, 6 - Block1: - Content-Format: 1544
14:53:43.613 ACK-2.04 10023 1ebdc8 Block1:
14:53:43.613 CON-PUT 10024 1ebdc8 Uri-Path: 5, 0, 6 - Block1: - Content-Format: 1544
14:53:43.743 ACK-2.04 10023 1ebdc8 Block1:
Nothing after this. Server stop sending packets.
Console output:
Feb 24, 2016 10:53:43 PM org.eclipse.californium.core.network.Matcher receiveResponse
INFO: Duplicate response for open exchange: ACK-2.04 MID=10023, Token=1ebdc8, OptionSet={"Block1":"(szx=5/512, m=false, num=336)"}, no payload
Feb 24, 2016 10:53:43 PM org.eclipse.californium.core.network.Matcher receiveResponse
WARNING: Possible MID reuse before lifetime end: 1ebdc8 expected MID 10024 but received 10023
Currently we have circumvented this problem by ignoring duplicate packets and not sending ACKs for duplicate packets.
Is that the correct approach ?
Is this a bug in Californium ? What is the correct approach/solution for this problem ?
Thanks
