One link of eth-trunk binding interface failure caused parts of IP address cannot ping

Publication Date:  2012-11-06 Views:  290 Downloads:  0
Issue Description
Topology:
NE40E-1-------S9312-1------S9312-2-------NE40E-2
Description: Two S9312 switches used 2 GE to interconnect by transmission wavelength division, access to eth-trunk binding interface, S9312 interface connected with 2 of NE40E, all the device start ospf protocol, and import static direct connection.
Failure:there is a direct network segment on S9312-2, NE40E-1 cannot ping some address of this network segment, NE40E-2 can ping all the IP address of this network segment.
Alarm Information
NULL
Handling Process

1. Check NE40E route table, it can learn the route of 9312-2 direct network segment.
2. 9312-1 can ping all the IP address of 9312-2 direct network segment.
3. Check the physic interface of 9312-1:
Eth-Trunk2                  up    up       0.01%  0.01%          0          0
  GigabitEthernet2/0/1      up    up          0%  0.01%          0          0
  GigabitEthernet3/0/1      up    up          0%     0%          0          0
4. Check the physic interface of 9312-2:
Eth-Trunk2                  up    up       0.01%  0.01%          0          0
  GigabitEthernet2/0/1      up    up          0%  0.01%          0          0
  GigabitEthernet3/0/1      down down         0%     0%          0          0
5. shutdown GigabitEthernet3/0/1 of 93120-1, failure disappeared.
Root Cause
NULL
Suggestions
Transmission wavelength division interface adopt the force mode, 9312 local side changed the interface to force, so if there is one link that problematic, the interface will be up for one end and down for another end.

Elements: huawei device traffic forwards through source and destination, and it remembered the forwarding patch, if it appeared the sigle through that would be problematic.
Suggestion: Wavelength division configuration linkage, when one end is down, another one will be down, or it changed to autonegotiation+linkage mode.

END