Hi. We still have the problem I've described before on the list, even though I've received no reply... The problem is, that SMS sendout is delayed by as much as 3 minutes using EMI2. This contrasts we had no delay using EMI_IP... The reason, on the surface, is the following:
Kannel EMI2 supports keepalive IF you have a username/password connection to the SMSC. However, the connection we have does not have username/password. And thus, Kannel ignores the keepalive setting. I assume that the protocol dictates this - msg 31 is intended to keep such a username-identified connection open, and simply does not apply without it. However, if I kuggle the code to send the 31 anyway (by forcing the timeout value to be != 0), it 'actually works' - it makes a syntax error because of the 31, but this keeps the connection alive... This actually keeps the line open as I want (so I get 'instant delivery'), but this is also clearly an unacceptable solution. I guess my question is 'Why it work previously with EMI_IP' ? Another version of the question is "why does it make that 'silly' lets-sleep-for-a-minute-before-we-try-to-deliver-what-he-just-sent' thing" ? I can repost the 'sleep' pattern we get on EMI2, if anyone didn't get it from the previous post.
