access --- leaf ---- spine
ADCampus 3.0 standard topology
Customer recently facing some issue with the Network unstable issue while doing the Microsoft Teams Call and having voice drop issue, also they found that all H3C Leaf having duplicate IP issue.
Analysis of the diagnostic information revealed a large number of ip conflicts on all leaf.
And these conflict ip addresses all correspond to different actual end devices.
The site is recovered by restarting the dhcp server, excluding the possibility of static ip terminals and loops in the same layer 2 network domain. Static ip terminals generally do not lead to a large number of terminal ip address conflicts, and will not be recovered with the restart of the dhcp server; ip address conflicts caused by loops will generate a large number of terminal ip conflicts. However, it will not be recovered with restarting the dhcp server, so the above two possibilities are excluded.
Go to the dhcp server backend to check the event logs and find that the primary and backup dhcp server synchronization timeout has triggered a failover
The dhcp server is in master/standby mode and the clock difference between the two exceeds 60 seconds, triggering a failover. At this point, the primary and backup dhcp servers lose contact, the backup dhcp server starts up, and the terminal comes online to obtain an ip address from one of the dhcp servers, which is unused in the other dhcp server and will still be assigned to other terminals, resulting in an address conflict and thus a bad address, and other ip conflicts in the same way.
The primary and secondary dhcp servers are configured with ntp servers to synchronise their clocks to avoid failover triggered by clock differences. Then restart the DHCP server.
Login before you can operate!login