memcached losing connections

[Kevin Dalley]

I am having trouble with memcached-1.2.4 on FreeBSD 4.1.

I'm using libmemcached as the client library.

Often, I receive a SIGPIPE on the client.  It appears that the socket
closes.  This seems more likely to happen when there are multiple
clients connected to the server.

Alternatively, when sending multiple set using non-blocking io, send
times out.  It appears that memcached loses track of the socket, and
forgets that there is outstanding data to be read.  This is with one
client.  The libmemcached test case "./testapp generate_nonblock"
reliably shows this problem.  The client is stuck in a poll, waiting for
the ability to write on a socket.

Has anyone seen this problem, preferably on FreeBSD, or a related OS,
but possibly on another machine.

When I run the same tests on libmemcached using a server on Linux, I do
not experience this problem.