Good day, Community!
This post explains the procedure to solving the ARP detect fail reason. Please see more details as you read below.
ISSUE DESCRIPTION
Cisco phones connected to a Huawei switch and have the MAC authentication from an Agile controller. The phones can authenticate successfully, but after some time, the phone is disconnected, showing 'Phone is registering' and it takes about 10 minutes to register again.
ISSUE ANALYSIS
1. Checked the switch diagnostics information.
2. From the switch, please check the following (Note: choose MAC address of the test phone):
[huawei]diagnose
[huawei-diagnose]dis aaa abnormal-offline-record mac XXXX-XXXX-XXXX
[huawei-diagnose]dis aaa offline-record mac XXXX-XXXX-XXXX
ROOT CAUSE
After checking the diagnostics of the switch and the aaa offline record logs for the IP phone, we found this:
[AD-E-GF-IDT1-SW1-diagnose]dis aaa offline-record mac XXXX-XXXX-XXXX
------------------------------------------------------------------------------
User name : c4729550f08e
Domain name : adnoc.sch.ae
User MAC : XXXX-XXXX-XXXX
User access type : MAC
User access interface : GigabitEthernet1/0/10
Qinq vlan/User vlan : 0/106
User IP address : 10.22.106.11
User IPV6 address : -
User ID : 661
User login time : 2019/02/18 17:54:11
User offline time : 2019/02/18 17:59:15
User offline reason : ARP detect fail
We found a lot of arp detect fail records from the provided logs and the user is offline after about 5 min from the login.
The device sends an ARP probe packet to check the user's online status. If the user does not respond within a detection period, the device considers that the user is offline.
If the VLAN to which the user belongs does not have a VLANIF interface or the VLANIF interface does not have an IP address, the device sends an offline detection packet using 255.255.255.255 as the source IP address. If a user cannot respond to an ARP probe packet with the source IP address 255.255.255.255, you can run the access-user arp-detect default ip-address command to specify a default source IP address for the offline detection packet.
To solve this issue, we need to set the default source IP address of offline probe packets to 0.0.0.0:
REFERENCE: