Re: network problems?

Robert Watson wrote:
> On Thu, 19 Apr 2007, Krassimir Slavchev wrote:
>
>> The problem is when I try to access ftp servers, the connection 
>> stalls randomly. Also I can't do cvsup and fetch. This happens only 
>> with machines running -current and when the traffic is passed through 
>> router based on FreeBSD 4.4. One of the test machines is my notebook 
>> which have installed 7.0-CURRENT (from today) and 5.4-STABLE and I 
>> see this problem only with -current.
>>
>> Is there any new features in -current tcp stack which may be 
>> incompatible with FreeBSD 4.x?
>
> No, not in principle, and ideally also not in practice.
>
> Sounds like a bit more diagnosis is needed, though.  The first thing 
> we should try to determine if this is a problem with a device driver, 
> the network stack, or applications.
>
> If you run ping on one terminal to the local router, does it 
> experience problems at the same time as other applications?  I.e., 
> does ftp stallage align with ping stallage?
>  
No, ping continues.
I have made tests with 'mrt' and nothing.
> Are there any console messages suggesting driver problems, such as 
> messages about interrupts, timeouts, and so on?
>
No, all seems to be normal.
> If you run "vmstat -i", is there any device with an extraordinarily 
> high interrupt rate (second column over 10000 or so)?
>
7.0-CURRENT on HP nx9010:

vmstat -i:

interrupt                          total       rate
irq0: clk                         226314        992
irq1: atkbd0                         650          2
irq6: fdc0                            11          0
irq8: rtc                          29037        127
irq9: acpi0                         2108          9
irq10: fwohci0+++                    456          2
irq12: psm0                         1501          6
irq14: ata0                         1863          8
irq15: ata1                           57          0
Total                             261997       1149

dmesg:
...
sis0: <NatSemi DP8381[56] 10/100BaseTX> port 0x2400-0x24ff mem 
0xd4008000-0xd4008fff irq 10 at device 18.0 on pci0
sis0: Silicon Revision: DP83816A
miibus0: <MII bus> on sis0
ukphy0: <Generic IEEE 802.3u media interface> PHY 0 on miibus0
ukphy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
sis0: Ethernet address: 00:0f:20:26:6e:78
sis0: [ITHREAD]
...

The 'vmstat -i' does not show anything about sis0!?

7.0-CURRENT on arm board with the same problem:

# vmstat -i
interrupt                       total       rate
irq1: at91_pio0                 25079        100
irq10: at91_mci0                70116        281
irq24: ate0                       218          0
irq13: at91_spi0                    6          0
Total                           95419        383

dmesg:
...
ate0: <EMAC> mem 0xdffbc000-0xdffbffff irq 24 on atmelarm0
miibus0: <MII bus> on ate0
rlphy0: <RTL8201L 10/100 media interface> PHY 16 on miibus0
rlphy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
ate0: Ethernet address: 00:ff:01:00:00:43
ate0: [ITHREAD]
...
> Could you try doing your network tests directly with the local router 
> and avoid using the wide area network?  This would help avoid having 
> wide area issues affect your testing, and also demonstrate whether or 
> not you can reproduce it in a purely local setup (much easier to debug).
>
Yes. There are problems with  HTTP traffic through this router too.
lynx dies with:

Sending HTTP request.
HTTP request sent; waiting for response.
Alert!: Unexpected network read error; connection aborted.
Can't Access `http://mysite.com/'
Alert!: Unable to access document.

Some pages loaded like apache server default page!?
Also icq works!?

May be it depends on the packet size but decreasing MTU to 1400 doesn't 
help.

I can't repeat the problem with my local ftp servers.

> Robert N M Watson
> Computer Laboratory
> University of Cambridge
>