Comment 2 for bug 880863

Thanks for the swift reply!

To test, I stopped the service, removed the database file (which I only just noticed) and started the service again. On the first connection I see it skip that block and thus use the value supplied in the connection call, which is correct.

In the next connection I see it find the client in the database and retrieve the correct value, so it does appear to end up in the database correctly as part of the first connection.

Therefore as I understand it, what I reported only manifests itself as a problem in the case where a client record is made and subsequently the requested keepalive changes. Incidentally I don't recall ever asking for a keepalive of 60, but my use of the client is still in early development, so it's possible. Alternatively it could be an artifact of my moving from the v0.12 server and not removing the database.