If I set up SQL Relay to use dynamic scaling, it eventually spawns off as many connections as it can and they never die off. Whats up with that?
This can happen if a spike of client traffic comes in. If configured to do so, SQL Relay will fork off more connection daemons to handle the traffic. Later when the traffic dies down, the connections that were forked off will still end up handling clients. Each connection daemon will experience longer idle periods than before the spike, but unless those idle periods are shorter than the ttl parameter, the connections will not die off. Setting the ttl shorter should take care of this, but it may take a lot of fine tuning to find the correct ttl. As of version 0.41, SQL Relay supports a new parameter: maxsessioncount. If it is set greater than 0, then connections that were forked off will only handle that many client sessions before dying off, even if it doesn’t go idle. Using this new parameter will ensure that all forked-off connections will eventually die off. Like the ttl, this parameter still requires some tuning but setting it too low will result in poorer performance rather than a
Related Questions
- Why doesn the last connection die off when I set connections=0 and maxconnections to some larger number and let SQL Relay go idle for a long time?
- If I set up SQL Relay to use dynamic scaling, it eventually spawns off as many connections as it can and they never die off. Whats up with that?
- Why Does My Wi-Fi Data Rate Keep Changing?
