Software guy here with a networks question. The problem is behind us, but I’d really like to know what happened.
For many years, our networks have run with IGMP snooping enabled. Once or twice a month, local applications on 30 to 40 clients are started that communicate with a suite of servers, using a few TCP connections, but mostly JOINing multicast traffic.
Three or four times over these years, IGMP handshaking stopped working. This last time, sometimes we could get 10 clients going, sometimes 30. Once clients started to fail to join, all previous clients would start having trouble making database connections, but would still receive their multicast data.
The networks group toggled a router’s IGMP snooping off and back on, and all was well again. Each time, this was the solution.
I‘m guessing there’s a connections table or some other cache that’s corrupted and the toggling resets the table.
Does anyone know what could be happening and why?
No comments:
Post a Comment