I have been working and doing some labs with GNS3; playing with SPT-threshold on PIM-SM, PIM-BIDIR and Anycast RP. I was running into quite a few issues where I would run a multicast source with a repeat 1000 and, after a couple of minutes, my IGP would start to flap. At first, I only noticed this problem because I was losing ICMP packets but after a debug I noticed this was because my IGP (in this case OSPF) was flapping for a few seconds. After searching for a solution online, I have found that this is a problem with GNS3 and caching.
To resolve this I had to disable cef with no ip cef on every router that was passing multicast traffic and also had to disable caching on the interfaces that were enable with PIM with no ip route-cache and no ip mroute-cache. After this, the issue was not causing any more route-flaps but I was still losing some multicast traffic. I believe this is because the mroute buffers clear out fast enough to not flap my IGP but not fast enough to not drop any traffic. The reason this whole issue was happening is because IGP’s use multicast to maintain adjacencies and when the cache was full and dumped, the adjacency were torn down.
If you are having weird issues with multicast try this fix. Of course, on real routers you wouldn’t need to do this.