The customer reported that ECMP routes failed to be delivered to the FIB table.
Checked routing and FIB information and found that multiple ECMP routes existed but the corresponding FIB entry contained only one next hop, indicating that ECMP routes failed to be
Checked whether the number of ECMP groups exceeded the maximum number, and found that only 15 ECMP groups existed.
Analyzed diagnostic logs and found that the ECMP routes failed to be delivered because chip resources were used up according to return values.
Analyzed the code and found that there was a possibility that ECMP resources failed to be released when ECMP routes flapped. Checked the log.log file and found that BGP flapped frequently, triggering ECMP route updates and ECMP resource release failures.
When ECMP routes flap, there is a possibility that ECMP resources are not released because of a bug in the code.
There is no reclaim mechanism for resources that fail to be released. The switch needs to be restarted to solve this problem. This bug has been fixed in V200R001.
Upgrade the system software to V200R001SPC700 and install the latest patch.