Re: Half-closed connections eat 100% cpu

Andres Kroonmaa wrote:

> Sometimes 0-byte reads are requested, not sure why, but they also cause
> clientReadRequest() do nothing, looks like half-closed, but skips defer.
> Maybe delay-pools are also involved here (0-byte reads?).

0-byte reads are bugs and need to be corrected. Would for sure cause
100% CPU as the socket would still be ready after reading nothing from a
ready socket..

(I assume you are not talking about the mandatory 0-byte read retsult
that occurs at end-of-file, which should be read once, or once per
second for half-closed sockets. None of these are 0-byte reads, only the
result is 0-byte...)

> Lately I saw cpu100% on sockets that shortly become idle serverside
> sockets. Kind of different problem.

Maybe not..

--
Henrik