| From peter.bentley@nomura.co.uk Fri Jun 19 10:01:21 1998
|
| Ken Wellsch wrote:
| > | On Jun 18, Ken Wellsch wrote
| > | > Within what seems like a few seconds I've consumed all my mbuf space,
| > | > crippling my server system.  netstat claims I've got the expected 16Kb
| > | > in the queue.
| > | >
| > | > The same server to another system, e.g. W95, W/NT, Linux does not have
| > | > this problem - the NetBSD seems to be properly blocking my writes.
| 
| Have you been able to verify (eg with truss) that your writes are being
| blocked? I know that the 16K in the send queue points to that, but if
| the data consuming the mbufs isn't coming from your app, and if the Mac
| isn't sending data to you then it must be triggering some very odd
| breakage inside the TCP stack, perhaps by advertising dumb window sizes.

Sorry if I implied that I am being blocked.  No I am definitely not
being blocked - that is the problem I think.  On other systems and
with other than Mac clients, I am being blocked as the pipeline fills
up.  But with NetBSD->Mac I don't get blocked and the kernel is being
fooled somehow into consuming all my mbuf's.  My server is just blindly
writing all that it can... relying on the kernel to block it.  The
Mac is slowly reading & processing the data while the mbuf's fill.

| > I have watched via "netstat -m" as the percent used number grows
| > rapidly when I'm running my application.  I mean rapidly too.  I
| > ran "netstat" as quickly as I could and the %-used increased something
| > like 6% to start, 23%, 56%, 73% and that is when I killed the process.
| 
| Perhaps if you could capture the start of the TCP stream with tcpdump,
| one of the TCP gooroos can spot what the problem is...

Yes, I can attempt to do that when I don't have an office worth of folks
also using the same network... 8-)

-- Ken