slowdown on primary network interface on news server... (2.2.10

Joel Jaeggli (joelja@darkwing.uoregon.edu)
Fri, 6 Aug 1999 10:29:10 -0700 (PDT)

I've got a news server that has recently become multi-homed. until now it
has behaved flawlessly for a few months, however now the interface that
has the default route slows to a crawl after running for a few hours, the
second interface is unaffected. no kernel messages accompany this behavior
however downing the interface and bring it back up again fix the problem.
generally it takes 8-11 hours for this issue to manifest itself.

sample pinges through each interface follow:

[news@platform articles]$ ping hammer <------ (this is over eth0)
PING hammer.uoregon.edu (128.223.220.29): 56 data bytes
64 bytes from 128.223.220.29: icmp_seq=0 ttl=255 time=1994.8 ms
64 bytes from 128.223.220.29: icmp_seq=1 ttl=255 time=2000.1 ms
64 bytes from 128.223.220.29: icmp_seq=2 ttl=255 time=2000.3 ms
64 bytes from 128.223.220.29: icmp_seq=3 ttl=255 time=2000.3 ms
64 bytes from 128.223.220.29: icmp_seq=4 ttl=255 time=1000.6 ms
64 bytes from 128.223.220.29: icmp_seq=5 ttl=255 time=56.3 ms
64 bytes from 128.223.220.29: icmp_seq=6 ttl=255 time=21.2 ms
64 bytes from 128.223.220.29: icmp_seq=7 ttl=255 time=624.8 ms
64 bytes from 128.223.220.29: icmp_seq=8 ttl=255 time=1035.0 ms

--- hammer.uoregon.edu ping statistics ---
10 packets transmitted, 9 packets received, 10% packet loss
round-trip min/avg/max = 21.2/1192.6/2000.3 ms

[news@platform articles]$ ping hammer-hpc.ogig.net <---- (this is over eth1)
PING hammer-hpc.ogig.net (198.32.163.71): 56 data bytes
64 bytes from 198.32.163.71: icmp_seq=0 ttl=255 time=0.2 ms
64 bytes from 198.32.163.71: icmp_seq=1 ttl=255 time=0.2 ms
64 bytes from 198.32.163.71: icmp_seq=2 ttl=255 time=0.2 ms
64 bytes from 198.32.163.71: icmp_seq=3 ttl=255 time=0.3 ms
64 bytes from 198.32.163.71: icmp_seq=4 ttl=255 time=0.3 ms

--- hammer-hpc.ogig.net ping statistics ---
5 packets transmitted, 5 packets received, 0% packet loss
round-trip min/avg/max = 0.2/0.2/0.3 ms
[news@platform articles]$

normally what you'd expect to see over eth0 is:

PING hammer.uoregon.edu (128.223.220.29): 56 data bytes
64 bytes from 128.223.220.29: icmp_seq=0 ttl=255 time=0.2 ms
64 bytes from 128.223.220.29: icmp_seq=1 ttl=255 time=0.2 ms
64 bytes from 128.223.220.29: icmp_seq=2 ttl=255 time=0.2 ms
64 bytes from 128.223.220.29: icmp_seq=3 ttl=255 time=0.2 ms

since there's just a intel fast ether switch between them.

hammer and hammer-hpc.ogig.net are two interfaces on the same remote
machine.

the machine is:

abit-bx6 2.0
intel p2-233
192mb ram
promise ultra-66 ide controller
2 x 9gb wd expert ata/66
2 x 18gb wd expert ata/66
smc etherpower II 10/100 (eth0)
bay networks ga-620 gigabit-sx gig ether (eth1)

the kernel is:

2.2.10

with the addition of:

2.2.10.uniform-ide-6.20.draco.patch
raid0145-19990724-2.2.10.gz

--------------------------------------------------------------------------
Joel Jaeggli joelja@darkwing.uoregon.edu
Academic User Services consult@gladstone.uoregon.edu
PGP Key Fingerprint: 1DE9 8FCA 51FB 4195 B42A 9C32 A30D 121E
--------------------------------------------------------------------------
It is clear that the arm of criticism cannot replace the criticism of
arms. Karl Marx -- Introduction to the critique of Hegel's Philosophy of
the right, 1843.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/