Hello All, I stumbled on this lately. We have a system where we use policy routing tables extensively. We treat them as use and throw, therefore we create them, install the needed routes and then when the job is done, we delete the route from the table and also the policy rule tied with the table. We use ip route flush table to clean the table and to delete the table as well. As expected, this leads us ending up creating huge chunk of tables on the system. Lately, its been found, that the tables that we flush appears as if they are not deleted. They are definitely empty, but aren’t deleted (ip route show table X, dumps empty, not the error dump terminated) and therefore exist in the hash map, leading to cpu spiking in fib_get_table(). Is this expected, or we doing doing something wrong here ? I probed fib_frontend.c and fib_trie.c in the kernel tree, and as far as I can see, it appears the tables are deleted only on netns exit. शुभ कामनाएँ, Varun Tewari
