SEEED Orin J4012 Network interface throuput issues

Products & Technology
#1

Hello,
We are trying to stream 2MBps stream to j4012 which is acting as RTSP client:
We receive many RX Drops and overflow errors.

No matter what ethernet port we select: eth0 or eth1
No matter if we have a switch between client and server or not.
We are using stock image jetpack5.1.1
What can be the source of the issue?
Thanks,
Oshi

We have made changes to buffer limits to support streaming:

net.ipv4.tcp_rmem = 4096 262144 16777216
net.ipv4.tcp_wmem = 4096 262144 16777216
net.core.netdev_max_backlog = 5000
net.core.wmem_max=12582912
net.core.rmem_max=12582912
net.ipv4.tcp_rmem= 10240 87380 12582912
net.ipv4.tcp_wmem= 10240 87380 12582912

Kernel Interface table

Iface     MTU   RX-OK RX-ERR RX-DRP RX-OVR   TX-OK TX-ERR TX-DRP TX-OVR Flg
can0       16       0     0     0 0            0     0     0     0 ORU
eth0     1500 48029671     0   290 0         4392     0     0     0 BMRU
eth1     1500  238740     0  3603 0         4295     0     0     0 BMRU
lo      65536    1413     0     0 0         1413     0     0     0 LRU
rndis0   1500       0     0     0 0            0     0     0     0 BMU
usb0     1500       0     0     0 0            0     0     0     0 BMU
wlan0    1500       0     0     0 0            0     0     0     0 BMU

Ip:
    Forwarding: 2
    48876911 total packets received
    11 with invalid addresses
    0 forwarded
    0 incoming packets discarded
    48876719 incoming packets delivered
    9829 requests sent out
    15776 dropped because of missing route
    14 reassemblies required
    7 packets reassembled ok
Icmp:
    1091 ICMP messages received
    37 input ICMP message failed
    ICMP input histogram:
        destination unreachable: 11
        echo requests: 931
        echo replies: 112
    1271 ICMP messages sent
    0 ICMP messages failed
    ICMP output histogram:
        destination unreachable: 116
        echo requests: 224
        echo replies: 931
IcmpMsg:
        InType0: 112
        InType3: 11
        InType8: 931
        InType9: 37
        OutType0: 931
        OutType3: 116
        OutType8: 224
Tcp:
    169 active connection openings
    3 passive connection openings
    2 failed connection attempts
    1 connection resets received
    6 connections established
    4979 segments received
    6469 segments sent out
    1 segments retransmitted
    0 bad segments received
    169 resets sent
Udp:
    46181585 packets received
    42863 packets to unknown port received
    2536334 packet receive errors
    2196 packets sent
    2536334 receive buffer errors
    0 send buffer errors
    IgnoredMulti: 33103
UdpLite:
TcpExt:
    145 TCP sockets finished time wait in fast timer
    31 delayed acks sent
    Quick ack mode was activated 1 times
    608 packet headers predicted
    433 acknowledgments not containing data payload received
    1383 predicted acknowledgments
    TCPSackRecovery: 1
    1 fast retransmits
    TCPDSACKOldSent: 1
    7 connections reset due to unexpected data
    TCPSackShiftFallback: 1
    IPReversePathFilter: 26
    TCPRcvCoalesce: 52
    TCPAutoCorking: 176
    TCPWantZeroWindowAdv: 1
    TCPOrigDataSent: 3980
    TCPDelivered: 4125
IpExt:
    InMcastPkts: 133096
    OutMcastPkts: 186
    InBcastPkts: 33110
    OutBcastPkts: 18
    InOctets: 69546439932
    OutOctets: 1054176
    InMcastOctets: 29942300
    OutMcastOctets: 78736
    InBcastOctets: 8949598
    OutBcastOctets: 1391
    InNoECTPkts: 48876951
#2

Hi there,
I would recommend , Enable Jumbo packet’s in the NIC see if that has any effect, obviously a flow control issue, does it just throw errors or is too slow with the retransmissions
Also if the nic is set for auto instead of FDX or HDX. what options do you have available on the nic?
(like right clicking on a Network interface 's hardware config in Windows)

NIC_Options
NIC_Options1627×1133 85.4 KB

HTH
GL :slight_smile: PJ

#3

Thanks for you reply,

However we think NIC is fine.
We see it using bmon. NIC receives all frames without packet loss. It looks like the problem is somewhere higher in th software stack.

image
image720×405 55.9 KB

#4

Hi there,
Ah’ Ok that is good to know. How much Ram you have on that J4012?
WHat is the system load like while doing it?
AFAIK, Linux sets the maximum value that can be configured for receive buffer, which can be viewed,

  • /proc/sys/net/core/rmem_default: Default receive buffer value used…
    typically Linux setting an initial value based on the memory size at startup.

What about slowing it (interface down) to a 10/100 speed?
Why so many ICMP’s , is the host connection stable? Lots of stuff happening.
? Looks right should work, buffers are pretty much all that’s left ?
HTH
GL :slight_smile: PJ

shot in the Dark , Try :v:
“sysctl -w net.core.rmem_max=26214400 # Set to 25M”