From: "Omar Polo" Subject: Re: relayd: reclassify log levels To: Rafael Sadowski Cc: tech@openbsd.org Date: Thu, 05 Nov 2026 22:03:16 +0100 Rafael Sadowski wrote: > OK? two questions but otherwise makes sense to me (but i'm not a relayd user :) > commit 7bdd895e0ef8ce3d5b2c8bbc4ec3a431cc454375 > Author: Rafael Sadowski > Date: Tue Nov 3 22:04:19 2026 +0100 > > relayd: reclassify log levels > > Some messages do not set errno, so they correspond to log_warnx rather > than warn. Some debug messages are actually warnings, others are simply > informational. > > diff --git a/ca.c b/ca.c > index beb86d5..a6e28b5 100644 > --- a/ca.c > +++ b/ca.c > @@ -259,7 +259,7 @@ ca_dispatch_relay(int fd, struct privsep_proc *p, struct imsg *imsg) > if (proc_composev_imsg(env->sc_ps, PROC_RELAY, > cko.cko_proc, imsg_get_type(imsg), -1, -1, iov, > c) == -1) > - log_warn("%s: proc_composev_imsg", __func__); > + log_warnx("%s: proc_composev_imsg", __func__); but imsg_compose() (which is the fallible thingy in proc_composev_imsg AFAIK) sets errno, no? (maybe not super useful but still.) > break; > } > > [...] > diff --git a/relay.c b/relay.c > index 1f28ccd..42a56a3 100644 > --- a/relay.c > +++ b/relay.c > [...] > @@ -1582,7 +1582,7 @@ relay_connect_retry(int fd, short sig, void *arg) > if (errno == ENFILE || errno == EMFILE) > log_warn("%s: session %d: deferring forward", > __func__, con->se_id); maybe delete this > - log_debug("%s: session %d: " > + log_warnx("%s: session %d: " > "forward failed: %s, %s", __func__, > con->se_id, strerror(errno), > con->se_retry ? "next retry" : "last retry"); if we're promoting this to a warn level? > @@ -1651,7 +1651,7 @@ relay_connect(struct rsession *con) > [...]