[ale] Interface dropping packets, how to fix safely?

[Dow Hurst]

A "sync;init6" fixes the problem immediately.  With hardly any traffic 
for a couple of days after the long transfer that caused the problem, it 
still was painfully erratic and slow.  The reboot fixed the problem and 
the interface is fast again.
Dow


James P. Kinney III wrote:

>Dow,
>
>This smells like a dying nic to me. Give it a rest to cool down
>thermally and see if it's performance improves. If it does, replace it.
>
>On Wed, 2004-08-18 at 19:18, Dow Hurst wrote:
>  
>
>>eth2: VIA VT6102 Rhine-II at 0x9000, 00:0e:a6:a2:f6:3d, IRQ 11.
>>eth2: MII PHY found at address 1, status 0x786d advertising 01e1 Link 43e1.
>>eth2: Setting full-duplex based on MII #1 link partner capability of 43e1.
>>
>>
>>I am seeing a number of dropped packets on this interface.  This is a 
>>remote machine so I have to be careful what I do.  Is there a known way 
>>to find out if the switch this interface is connected to is running 
>>half-duplex?  Can I just find out the driver flags to switch it to half 
>>duplex without too much risk?
>>
>>Initially after boot there wasn't any apparent problems, however, after 
>>a long upload of a lot of data the interface is almost unuseable with 
>>long delays between letters typed while using SSH.
>>
>>Note this later /var/log/messages entry:
>>eth2: Setting half-duplex based on MII #1 link partner capability of 0000.
>>eth2: Setting full-duplex based on MII #1 link partner capability of 43e1.
>>
>>eth2      Link encap:Ethernet  HWaddr 00:0E:A6:A2:F6:3D
>>          inet addr:152.13.214.246  Bcast:152.13.215.255  Mask:255.255.254.0
>>          inet6 addr: fe80::20e:a6ff:fea2:f63d/64 Scope:Link
>>          UP BROADCAST NOTRAILERS RUNNING MULTICAST  MTU:1500  Metric:1
>>          RX packets:2138379 errors:0 dropped:19647 overruns:0 frame:0
>>          TX packets:1343549 errors:0 dropped:0 overruns:0 carrier:0
>>          collisions:0 txqueuelen:1000
>>          RX bytes:2265150739 (2160.2 Mb)  TX bytes:128588389 (122.6 Mb)
>>          Interrupt:11 Base address:0x9000
>>
>>Running mtr to track the route and see where the delays are shows that 
>>the last interface, namely this interface, is the big delay with an 
>>order of magnitude higher time than any of the hops between me and the 
>>machine.  I'm still in Acworth, GA while the server is at UNCG 
>>Greensboro NC.
>>
>>Thanks for your help!
>>Dow
>>
>>_______________________________________________
>>Ale mailing list
>>Ale at ale.org
>>http://www.ale.org/mailman/listinfo/ale
>>
>>!DSPAM:4123e3d721594769117387!
>>    
>>
>>------------------------------------------------------------------------
>>
>>_______________________________________________
>>Ale mailing list
>>Ale at ale.org
>>http://www.ale.org/mailman/listinfo/ale
>>