> However, I'd be glad if you could tell me why my writes block that > long. I'm certainly under 128MB of socket buffers, so the limits > (on my 1GB machine) should not be reached. > > How certain are you? 99% - unless the kernel counts unused timewait slab entries against the allocations. I tested with an otherwise idle machine, using 1000 connections of 14kB each, and the writes are degraded from the start. That's nowhere near 128MB (32k * PAGE_SIZE). > Why don't you add debugging to the kernel > so that you know it as a fact? The box is 30 min away. :-) > Come back when this experiment completes ;-) Will do. Patrick - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org
