Subject: Re: Are tlp[01] weak for big traffic?
To: None <port-cobalt@netbsd.org>
From: NAKAJI Hiroyuki <nakaji@jp.freebsd.org>
List: port-cobalt
Date: 12/20/2006 00:40:23
>>>>> In <87zm9l9yfr.fsf@roddy.4407.kankyo-u.ac.jp>
>>>>> NAKAJI Hiroyuki <nakaji@jp.freebsd.org> wrote:
> > > And, "ifconfig tlp0 down; sleep 10; ifconfig tlp0 up" does not help.
> > Why don't you try to get any statistics/logs (vmstat -i, netstat -i,
> > ipfstat, ipnat -s, ps, top, dmesg, tcpdump, /var/log/messages etc.) there?
> Thanks. I did not know much about these check tools, especially
> {vm,net}stat -i and ipnat -s. I'll try them.
> Hmm. dmesg and /var/log/messages don't have usuful messages about tlp or
> some network troubles.
I started the download of Solaris 10 (11/06) CD images whose size is
about 2400MB from sun.com this morning. And I found the transfer rate
became 0byte/s about 10 p.m. JST. Mrtg graph shows that the
input/output of tlp[01] keep around 30kB/sec (including ssh session).
Now, the network of RaQ2 is almost down. And here are some commands
outputs.
$ netstat -in
Name Mtu Network Address Ipkts Ierrs Opkts Oerrs Colls
tlp0 1500 <Link> 00:10:e0:00:6f:33 4764468 0 3819144 0 0
tlp0 1500 default 60.32.13.193 4764468 0 3819144 0 0
tlp0 1500 default fe80::210:e0ff:fe 4764468 0 3819144 0 0
tlp0 1500 default 2001:3e0:a84::2 4764468 0 3819144 0 0
tlp1 1500 <Link> 00:10:e0:00:6f:32 2030795 0 2951779 0 0
tlp1 1500 default 192.168.1.1 2030795 0 2951779 0 0
tlp1 1500 default fe80::210:e0ff:fe 2030795 0 2951779 0 0
tlp1 1500 default 2001:3e0:a84:1::1 2030795 0 2951779 0 0
lo0 33192 <Link> 2336395 0 2336395 0 0
lo0 33192 default 127.0.0.1 2336395 0 2336395 0 0
lo0 33192 default ::1 2336395 0 2336395 0 0
lo0 33192 default fe80::1 2336395 0 2336395 0 0
$ vmstat -i
interrupt total rate
soft serial 690003 3
soft net 5995407 28
soft clock 884734 4
cpu int 3 699386 3
cpu int 1 7475127 36
icu irq 14 1999722 9
cpu int 2 4845941 23
mips int 5 (clock) 20717858 100
Total 43308178 209
$ sudo ipfstat
bad packets: in 0 out 0
IPv6 packets: in 3856282 out 4119124
input packets: blocked 0 passed 8846359 nomatch 4149286 counted 0 short 0
output packets: blocked 0 passed 9102410 nomatch 4322852 counted 0 short 0
input packets logged: blocked 0 passed 0
output packets logged: blocked 0 passed 0
packets logged: input 0 output 0
log failures: input 0 output 0
fragment state(in): kept 0 lost 0 not fragmented 0
fragment state(out): kept 0 lost 0 not fragmented 0
packet state(in): kept 0 lost 0
packet state(out): kept 0 lost 0
ICMP replies: 0 TCP RSTs sent: 0
Invalid source(in): 0
Result cache hits(in): 4697073 (out): 4779558
IN Pullups succeeded: 36 failed: 0
OUT Pullups succeeded: 2716 failed: 0
Fastroute successes: 0 failures: 0
TCP cksum fails(in): 0 (out): 0
IPF Ticks: 414498
Packet log flags set: (0)
none
$ sudo ipnat -s
mapped in 2939912 out 2027595
added 2848 expired 0
no memory 0 bad nat 0
inuse 3
rules 3
wilds 0
I was running "tcpdump -n -i tlp0", "tcpdump -n -i tlp1" and "vmstat
-i -w 5" while downloading. The maximum rate of "soft net" was 29.
## Oops, why don't I use script to log all outputs...
And, tcpdump shows only SNMP, NTP, and SMB now. Dmesg and
/var/log/messages have no useful contents.
I suspected the local proxy (squid) performance or restriction, but no
doubt, because another Solaris download from FreeBSD/pc98 box on the
same segment with the proxy was fast enough and finished successfuly
in about an hour. This shows that ISP does not restrict the band width
nor max traffic.
Thanks.
--
NAKAJI Hiroyuki