anyone ever had issues with the routes table hanging? it seems to be stuck in the

NLMSG_OK

loop in https://github.com/osquery/osquery/blob/613028a6384d4e0b101fe087717c2dfdcde3efbf/osquery/tables/networking/linux/routes.cpp#L241 as I can see all the other stuff happening during a

strace

. it receives a message once and then gets stuck in a recv/sleep loop with no data being passed. not sure if there's a better channel than this one for general osquery debug stuff.

One of our customers hit this and we ended up reimplementing the retry/timeout here. I’ll post a PR tomorrow.

Thanks Ryan!

thank you ryan! i'm sure you can't share too many details, but i'd imagine the root cause for most folks here is docker (seemingly) creating 4 different v6 routes for every container leading to an extremely large v6 routing table.

The box we saw this on had 96 cores and 384GB of memory. Needless to say it had a lot of containers running on it :-)

lol we have almost exactly the same SKU. 🙂 the actual number seems to be around 15 when the table gets sad