After the restart of a 3650 distribution switch stack that is acting as an RP, a downstream switch was not able to send multicast traffic; other parts of the network were not affected. After considerable troubleshooting, we discovered that removing the register-source interface and re-adding immediately caused all multicast traffic to once again work. I guess that this caused a re-registration with the RP. Before this was performed all show commands suggested that PIM (sparse-mode) was configured and operating correctly and was currently registered with the RP. These switches are 3650 running IOS 16.X code. This is the second time this month that the switch required a reboot, and on both occasions the same issue happened with the same switch.
I would like to understand:
- What caused this problem and why multicast operation did not resume when the upstream RP switch rebooted.
- How it can be prevented in the event that the distribution switch needs restarted again.
If there are commands that would be beneficial to gather output from, I can put the network back in to a fault condition some weekend evening.
The commands to fix this fault were:
switch#sh run | i ip pim register-source
ip pim register-source Loopback0
switch#conf t
switch(config)#no ip pim register-source Loopback0
switch(config)#ip pim register-source Loopback0
Thank you.
