Hi,
pfsync does "bulk update" just after boot, I noticed it sometimes
fails. When finishing "bulk update", the duration in the "bulk-end"
packet and our duration based on uptime are compared, but that
comparision should be fixed. It must consider the values are rounded
in a second.
ok?
Consider being rounded in a second when comparing the duration in
"bulk-end" packet and the duration based on our uptime. This fixes
the problem the carp demote count sometimes becomes 33 after reboot.
Index: sys/net/if_pfsync.c
===================================================================
RCS file: /cvs/src/sys/net/if_pfsync.c,v
retrieving revision 1.274
diff -u -p -r1.274 if_pfsync.c
--- sys/net/if_pfsync.c 10 Jul 2020 13:26:42 -0000 1.274
+++ sys/net/if_pfsync.c 25 Jul 2020 05:09:47 -0000
@@ -1169,8 +1169,7 @@ pfsync_in_bus(caddr_t buf, int len, int
break;
case PFSYNC_BUS_END:
- if (getuptime() - ntohl(bus->endtime) >=
- sc->sc_ureq_sent) {
+ if (ntohl(bus->endtime) <= getuptime() + 1 - sc->sc_ureq_sent) {
/* that's it, we're happy */
sc->sc_ureq_sent = 0;
sc->sc_bulk_tries = 0;
