Claudio Jeker(cje...@diehard.n-r-g.com) on 2021.05.14 11:21:11 +0200:
> I think it is a good idea to limit the number of concurrent connects in
> bgpd. I used 32 as the limit since that is way enough for the number of
> RTR sessions people will configure.
>
> If the limit is hit the request will be dropped and the rtr process will
> retry the connect after the retry timeout. Hopefully by then the number of
> connections is down again.
ok benno
but i suggest to s/inflight/concurrent/, and keep the word inflight for log
messages about tracking connections for purpose of not running out of file
descriptors?
> --
> :wq Claudio
>
> Index: bgpd.c
> ===================================================================
> RCS file: /cvs/src/usr.sbin/bgpd/bgpd.c,v
> retrieving revision 1.236
> diff -u -p -r1.236 bgpd.c
> --- bgpd.c 11 May 2021 07:57:24 -0000 1.236
> +++ bgpd.c 11 May 2021 08:00:25 -0000
> @@ -74,6 +74,7 @@ struct connect_elm {
> TAILQ_HEAD( ,connect_elm) connect_queue = \
> TAILQ_HEAD_INITIALIZER(connect_queue);
> u_int connect_cnt;
> +#define MAX_CONNECT_CNT 32
>
> void
> sighdlr(int sig)
> @@ -1303,6 +1304,12 @@ bgpd_rtr_connect(struct rtr_config *r)
> struct connect_elm *ce;
> struct sockaddr *sa;
> socklen_t len;
> +
> + if (connect_cnt >= MAX_CONNECT_CNT) {
> + log_warnx("rtr %s: too many inflight connection requests",
> + r->descr);
> + return;
> + }
>
> if ((ce = calloc(1, sizeof(*ce))) == NULL) {
> log_warn("rtr %s", r->descr);
>
