Symptom
Poap using managment port over v6 network will not be able to download poap script.
Conditions
1. POAP is using v6 network.
2. Management port is getting used for POAP v6.
3. ipv6 address should be present over mgmt0 before triggering poap using "write erase" and "reload".
4. ipv6 address configured over mgmt0 should not be able to reach tftp server which gets assigned during DHCPv6 phase of poapv6.
5. Step 4 tells about the scenario where customers v6 mgmt0 address is in different network and DHCPv6 and TFTPv6 server is in different network.
Workaround
1st work around : Remove ipv6 address from mgmt0 before triggering poap.
2nd work around : ipv6 address on mgmt0 and ipv6 address assigned during poap both are in same network and able to access tftp server.
Further Problem Description
This problem is seen when we end up in a scenario where mgmt0 interface is having 2 ipv6 address configured during poap and
POAP tries to copy script from tftp server with any one of the v6 address (which is not reachable to tftp server).
below are the logs from poap v6 console session.
===============================
Using DHCP, valid information received over mgmt0 from fe80:0:0:0:4cec:74be:6c12:a137
Assigned IP address: 2001:db8:0:2:0:0:0:17d ===> v6 ip address provided by dhcp v6 server
Prefix: 2001:db8:0:2:0:0:0:0/64
DNS Server: 2001:db8:0:2:0:0:0:253
Next hop: 2001:db8:0:2:0:0:0:254
Bootfile url: tftp://2001:db8:0:2::253/poap3.py
The POAP Script download has started
The POAP Script is being downloaded from [copy tftp://2001:db8:0:2::253/poap3.py bootflash:scripts/script.sh vrf management ] ========> packet capture "poap-tftp-fail.pcap" shows that for tftp copy request 2001:xx::xx367:24 (masking public ipv6 address for security reasons, please refer source_ip_used_in_failed_tftp_copy.png to see actual ipv6 address )is used as source ip instead of 2001:db8:0:2:0:0:0:17d .
POAP boot file download failed.
POAP DHCP discover phase failed
