Sunday, January 22, 2012, 5:36:23 PM, you wrote: > As for accept threads, my reading of the code is that if frequent_accept is > false, then every normal IO thread also does accepts (via > NetAccept::init_accept_per_thread).
No, that's wrong. It calls NetAccept::init_accept and starts a single accept thread. Interesting, NetAccept::init_accept claims to be suitable for rapid accept sockets. If you run the regression, it will automatically enable tags during HttpTxnCache which demonstrate the problem. I will be removing that before I actually commit the patch, but it's handy for testing.
