discussion
[Top] [All Lists]

[Discussion] Re: Reduce polling costs

To: l.andrew@xxxxxxxx
Subject: [Discussion] Re: Reduce polling costs
From: Tom Quetchenbach <quetchen@xxxxxxxxxxx>
Date: Tue, 14 Aug 2007 10:22:43 -0700
Lachlan Andrew wrote:
Greetings all,

We're using web100 to monitor flows continuously. When we use the
libraries to poll web100 at intervals of 50ms or less, the overhead
from the fopen seems to become very significant, taking about as
much CPU as the ~1Gbps iperf process.

The problem seems to have to do with the fact that web100 locks the socket in connection_file_rw before reading. If we remove the calls to lock_sock and release_sock, the problem goes away.


So, is it dangerous not to lock the socket on a read? What if we call lock_sock as usual unless the connection state is TCP_ESTABLISHED--would that be sufficient protection against the connection disappearing while we're trying to read its web100 variables?

Thanks
-Tom

_______________________________________________
Discussion mailing list
Discussion@xxxxxxxxxx
http://internal.web100.org/mailman/listinfo/discussion

<Prev in Thread] Current Thread [Next in Thread>