-
Hi I was wondering if anyone had any suggestions on how to troubleshoot: "mroute: pending queue full, dropping entries" It seems, to me, to be related to the kernel asking mrouted for cache information. OS: Ubuntu 18.04.5 I've run the mrouted daemons with debug=cache, but I can't see any output that I can correlate to when the errors occur. Going to a VM with 64GB seemed to reduce how often the error is reported. The error seems to come from the kernel, source file: /linux/net/ipv4/ipmr.c
Thanks |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 2 replies
-
Sounds like you have a LOT of multicast coming in to your router. It's possible you see this because mrouted is not quick enough to process the IGMPMSG_NOCACHE messages from the kernel. This message is sent on the mutlicast routing socket for any new unknown multicast (S,G) pair. If the routing daemon knows of a receiver on another LAN that want this it creates an mroute for the corresponding outbound interface, if not then it takes note of the (S,G) pair and creates a "block" rule to prevent further notifications from the kernel. Such block rules have no outbound interfaces. The function responsible for creating all mroutes is |
Beta Was this translation helpful? Give feedback.
Sounds like you have a LOT of multicast coming in to your router. It's possible you see this because mrouted is not quick enough to process the IGMPMSG_NOCACHE messages from the kernel. This message is sent on the mutlicast routing socket for any new unknown multicast (S,G) pair. If the routing daemon knows of a receiver on another LAN that want this it creates an mroute for the corresponding outbound interface, if not then it takes note of the (S,G) pair and creates a "block" rule to prevent further notifications from the kernel. Such block rules have no outbound interfaces.
The function responsible for creating all mroutes is
k_add_rg()
, called fromprune.c:add_table_entry()
. You can ad…