Re: [Bugme-new] [Bug 11752] New: Extremely low netperf UDP_RR throughput for nvidia MCP65

[Rick Jones <rick.jones2@hp.com>]

Arno J. Klaassen wrote:
> Hello,
> 
> 
>>Are netstat stats showing retransmissions during a TCP_STREAM test?
> 
> 
> some more info :
> 
>   [root@localhost mcp65]# uname -a
>   Linux localhost.localdomain 2.6.27-0.352.rc7.git1.fc10.x86_64 #1 SMP Tue Sep 23 21:13:29 EDT 2008 x86_64 x86_64 x86_64 GNU/Linux                                
> 
> 
>   [root@localhost ~]# ifconfig
>   eth0      Link encap:Ethernet  HWaddr 00:1E:68:XXX
>           inet addr:172.16.1.31  Bcast:172.16.1.255  Mask:255.255.255.0
>           inet6 addr: XXX/64 Scope:Link           
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1           
>           RX packets:1731 errors:149 dropped:0 overruns:0 frame:149    
>           TX packets:1628 errors:0 dropped:0 overruns:0 carrier:0      
>           collisions:0 txqueuelen:1000                                 
>           RX bytes:2549871 (2.4 MiB)  TX bytes:125378 (122.4 KiB)      
>           Interrupt:20 Base address:0x6000
> 
> After some fiddling (essentially installing a netperf-rpm) :
> 
>   [root@localhost mcp65]# netstat -Ieth0                                         
>   Kernel Interface table                                                         
>   Iface       MTU Met    RX-OK RX-ERR RX-DRP RX-OVR    TX-OK TX-ERR TX-DRP TX-OVR Flg                                                                             
>   eth0       1500   0     6347    508      0      0     5838      0      0      0 BMRU                                                                            
> 
> 
>   [root@localhost mcp65]# ethtool eth0                                            
>   Settings for eth0:                                                              
>         Supported ports: [ MII ]                                                
>         Supported link modes:   10baseT/Half 10baseT/Full                       
>                                 100baseT/Half 100baseT/Full                     
>                                 1000baseT/Full                                  
>         Supports auto-negotiation: Yes                                          
>         Advertised link modes:  10baseT/Half 10baseT/Full                       
>                                 100baseT/Half 100baseT/Full                     
>                                 1000baseT/Full                                  
>         Advertised auto-negotiation: Yes                                        
>         Speed: 1000Mb/s                                                         
>         Duplex: Full                                                            
>         Port: MII                                                               
>         PHYAD: 1                                                                
>         Transceiver: external                                                   
>         Auto-negotiation: on                                                    
>         Supports Wake-on: g                                                     
>         Wake-on: d                                                              
>         Link detected: yes                                                      
> 
>   [root@localhost mcp65]# netperf -v -t TCP_STREAM -H 172.16.1.7
>   TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 172.16.1.7 (172.16.1.7) port 0 AF_INET
>    132.64

Hmm, I'm surprised that the lack of a value  following the -v was 
successful - I'll have to go back and look at the code :)  Still, I 
guess it gave you the desired "-v 0" behaviour.

>   [root@localhost mcp65]# netstat -Ieth0
>   Kernel Interface table
>   Iface       MTU Met    RX-OK RX-ERR RX-DRP RX-OVR    TX-OK TX-ERR TX-DRP TX-OVRFlg
>   eth0       1500   0    84682   2933      0      0   127298      0      0      0BMRU
> 
> 
> I hope this is what you asked for ...

Close enough.  I suspect that if you were to snap netstat -s -t before 
and after the netperf you'd have seen retransmissions correlated to 
those "RX-ERR" stats.  My history is such that I don't think of netstat 
for link-level stats and only think of it in the context of IP-layer and 
higher (eg tcp).

Clearly something is fubar with the rx side (well duh :).  The next set 
of stats I'd try to look at would be ethtool stats for the interface, eg 
ethtool -S eth0 and see if it shows someting more specific for the 
"RX-ERR" shown by netstat -I eth0.

rick jones