Hi all, We're using Kannel to connect to an SMSC through SMPP 3.4. While testing with the sctt SMSC simulator, we sometimes see the following: Kannel submits an SMS message (bearerbox log, log-level 0, with line numbers): [...] 1555 2009-03-31 10:16:05 [25853] [6] DEBUG: SMPP PDU 100512a40 dump: 1556 2009-03-31 10:16:05 [25853] [6] DEBUG: type_name: submit_sm 1557 2009-03-31 10:16:05 [25853] [6] DEBUG: command_id: 4 = 0x00000004 1558 2009-03-31 10:16:05 [25853] [6] DEBUG: command_status: 0 = 0x00000000 1559 2009-03-31 10:16:05 [25853] [6] DEBUG: sequence_number: 33 = 0x00000021 1560 2009-03-31 10:16:05 [25853] [6] DEBUG: service_type: NULL [...] Bearerbox submits an SMS message to the SMSC simulator with the sequence number 0x00000021 (line 1559) A fraction of second later, the SMSC simulator responds: [...] 1665 2009-03-31 10:16:05 [25853] [6] DEBUG: SMPP[sctt]: Got PDU: 1666 2009-03-31 10:16:05 [25853] [6] DEBUG: SMPP PDU 100509910 dump: 1667 2009-03-31 10:16:05 [25853] [6] DEBUG: type_name: submit_sm_resp 1668 2009-03-31 10:16:05 [25853] [6] DEBUG: command_id: 2147483652 = 0x80000004 1669 2009-03-31 10:16:05 [25853] [6] DEBUG: command_status: 0 = 0x00000000 1670 2009-03-31 10:16:05 [25853] [6] DEBUG: sequence_number: 33 = 0x00000021 1671 2009-03-31 10:16:05 [25853] [6] DEBUG: message_id: "00000022" 1672 2009-03-31 10:16:05 [25853] [6] DEBUG: SMPP PDU dump ends. 1673 2009-03-31 10:16:05 [25853] [6] WARNING: SMPP[sctt]: SMSC sent submit_sm_resp with wrong sequence number 0x00000021 [...] Note now that bearerbox does not recognize the sequence number 0x00000021 that it just submitted to the SMSC (line 1673). The end result is that Kannel re-submits this SMS message after 60 seconds, meaning that the end-user will have received this SMS message twice. This behavior occurs occasionally, in a seemingly random way. It seems to happen most often when bearerbox submits a whole bunch of SMS messages in a very short time. We observe it mostly when we reconnect the SMSC simulator after the connection was broken, and a bunch of SMS messages where queued in Kannel in the mean time. Also note that the message above was the last of a bunch of 20 SMS messages, but this one was the only one for which the sequence number was not recognized by Kannel. In some other tests, we got two or three unrecognized sequence numbers out of 20, in other tests all 20 sequence numbers were recognized. Has anyone observed this before? Any ideas on what could cause this? Wrong configuration, Kannel bug,...? Any ideas on what could solve this? Thanks, Raf
