Re: [ircu-development] Slow on high-MTU (local host) connections?

Carlo Wood (carlo@alinoe.com)
Wed, 7 Nov 2001 04:35:54 +0100


The examples in the previous mail are from the case with MTU is 8000
(and were to only two occurances of a EAGAIN for read() actually).

Allow me show the statistics for both MTU's:

MTU 16436:

~>grep -B1 '\[srvx2\] select.*<0\.[1-9]' mtu16436 | grep '\[srvx2\] read' | wc --lines
323
~>grep -B1 '\[srvx2\] select.*<0\.[1-9]' mtu16436 | grep '\[srvx2\] read.*EAGAIN' | wc --lines
323

Conclusion: ALL calls to select() that took longer than 0.1 second
were following a call to read() that failed with EAGAIN. In total 323 times.

MTU 8000:

~>grep -B1 '\[srvx2\] select.*<0\.[1-9]' mtu8000 | grep '\[srvx2\] read' | wc --lines
2
~>grep -B1 '\[srvx2\] select.*<0\.[1-9]' mtu8000 | grep '\[srvx2\] read.*EAGAIN' | wc --lines
2

Idem, but only two occurences.

The total number of calls to select in both are respectively:

~>grep '\[srvx2\] select.*' mtu16436 | wc --lines
1221
~>grep '\[srvx2\] select.*' mtu8000 | wc --lines
658

-- 
Carlo Wood <carlo@alinoe.com>

(Also forwarded because first I used a wrong address) - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/