most time, we always can see the outbound discard counter at some ports(like the below outputs).
10GE1/0/10 current state : UP (ifindex: 14)
Line protocol current state : UP
Description:
Switch Port, TPID : 8100(Hex), The Maximum Frame Length is 9216
IP Sending Frames' Format is PKTFMT_ETHNT_2, Hardware address is 80d4-a557-3101
Port Mode: COMMON COPPER, Port Split: -
Speed: 1000, Loopback: NONE
Duplex: FULL, Negotiation: ENABLE
Input Flow-control: DISABLE, Output Flow-control: DISABLE
Mdi: AUTO
Last physical up time : 2018-03-15 01:53
Last physical down time : 2018-03-15 01:53
Current system time: 2018-03-15 01:53
Statistics last cleared:2018-03-15 01:53
Last 300 seconds input rate: 1414214 bits/sec, 975 packets/sec
Last 300 seconds output rate: 3473927 bits/sec, 1579 packets/sec
Input peak rate 748508826 bits/sec, Record time: 2018-03-15 01:53
Output peak rate 964162432 bits/sec, Record time: 2018-03-15 01:53
Input : 4154362094372 bytes, 9244205266 packets
Output: 18809677741840 bytes, 22806069402 packets
Input:
Unicast: 9243209099, Multicast: 5034
Broadcast: 991133, Jumbo: 0
Discard: 0, Frames: 0
Pause: 0
Total Error: 0
CRC: 0, Giants: 0
Jabbers: 0, Fragments: 0
Runts: 0, DropEvents: 0
Alignments: 0, Symbols: 0
Ignoreds: 0
Output:
Unicast: 22701615452, Multicast: 22411756
Broadcast: 82042194, Jumbo: 0
Discard: 12153, Buffers Purged: 0
Pause: 0
we'd like to know the reason for the outbound discard. Here I'd like share the exact explanation for this issue.
First, we'd like to clarify that all these issues were caused by the micro burst traffic. The micro burst can cause congestion and further if the burst exceeds the outbound qos buffer, it will cause the packet drop at the outbound interface.<?xml:namespace prefix = "o" ns = "urn:schemas-microsoft-com:office:office" />
We can use below commands to check the outbound queue buffer and if the outbound discard happens:
[~huawei-diagnose]disp qos queue statistics interface 10ge 1/0/48 //this command can show us the last time when outbound discard happened.
Queue CIR/PIR Passed Pass Rate Dropped Drop Rate Drop Time
(kbps) (Packets/Bytes) (pps/bps) (Packets/Bytes) (pps/bps)
----------------------------------------------------------------------------------------------
0 0 15865481750 1408 19038 0 2018-03-23
10000000 39401771293195 48037712 97070350 0 10:45:08
----------------------------------------------------------------------------------------------
1 0 0 0 0 0 -
10000000 0 0 0 0
----------------------------------------------------------------------------------------------
2 0 0 0 0 0 -
10000000 0 0 0 0
----------------------------------------------------------------------------------------------
3 0 0 0 0 0 -
10000000 0 0 0 0
----------------------------------------------------------------------------------------------
4 0 0 0 0 0 -
10000000 0 0 0 0
----------------------------------------------------------------------------------------------
5 0 132403 0 0 0 -
10000000 48856707 96 0 0
----------------------------------------------------------------------------------------------
6 0 3 0 0 0 -
10000000 222 0 0 0
----------------------------------------------------------------------------------------------
7 0 1984230 0 0 0 -
10000000 126990720 256 0 0
----------------------------------------------------------------------------------------------
[~ERGSWICER01-diagnose]disp qos buffer-usage interface 10ge1/0/48 //this command can show us the queue buffer usage, if the queue buffer has the “Current” value, this means that the congestion is happening.
Total : 6784 cells (1378 KBytes)
Current used : 0 cells (0 KBytes)
Remained : 6784 cells (1378 KBytes)
Peak used : 5127 cells (1041 KBytes)
Average used : 0 cells (0 KBytes)
Buffer Usage on each Queue: (cells/KBytes)
-----------------------------------------------------------------------------
QueueIndex Current Peak Average
-----------------------------------------------------------------------------
0 0/0 5127/1041 0/0
1 0/0 0/0 0/0
2 0/0 0/0 0/0
3 0/0 0/0 0/0
4 0/0 0/0 0/0
5 0/0 0/0 0/0
6 0/0 0/0 0/0
7 0/0 0/0 0/0
-----------------------------------------------------------------------------
On CE switch, we have a command to set the qos burst-mode, so that some specific interface can get more buffer when the congestion happens, this command can release the congestion but couldn’t eliminate the congestion.
You can configure enhanced mode for the interfaces that have outbound discard counters, but this command is just to allocate more buffer for the specific interface, if the burst traffic is still high, the outbound packet drop may still persist.
[~HUAWEI-10GE1/0/1] qos burst-mode enhanced
From group: Switch
