On Thu, Jun 17, 2021 at 12:24:16PM +0200, Claudio Jeker wrote:
> On Mon, Jun 14, 2021 at 05:10:07PM +0200, Claudio Jeker wrote:
> > On Thu, May 27, 2021 at 06:24:06PM +0200, Claudio Jeker wrote:
> > > Implement RFC 7313 enhanced route refresh.
> > >
> > > While there also change when graceful restart EoR markers are sent.
> > > In short the graceful restart marker should only be sent initally. After
> > > that the End of Route Refresh message should be sent instead.
> > > Because of this track if an EoR marker was received or should be sent in
> > > the peer config.
> > >
> > > For now this setting is off by default but that may be changed at a later
> > > state.
> > >
> > > Please try this out and tell me if it works for you. The message and
> > > prefix/rrefresh counters in bgpctl show nei output help a lot to see what
> > > is going on.
>
> Minor cleanup, instead of this comment use CTASSERT to make sure that the
> recv_eor and send_eor fields are large enough to work as a bitfield for
> AID_MAX
>
> > -#define AID_MAX 5
> > +#define AID_MAX 5 /* check rde_peer.recv_eor when max
> > reaches 7 */
>
> Enhanced route refresh is currently off by default. So unless somebody
> is against this I will commit this soon. At least then it gets tested :)
Under what circumstances will bgpd(8) send an enhanced route refresh?
If I issue 'bgpctl neighbor xxx refresh' on a session where Enhanced
Refresh is negotiated
Negotiated capabilities:
Multiprotocol extensions: IPv4 unicast
4-byte AS numbers
Route Refresh
Enhanced Route Refresh
Graceful Restart
I see in tshark:
Border Gateway Protocol - ROUTE-REFRESH Message
Marker: ffffffffffffffffffffffffffffffff
Length: 23
Type: ROUTE-REFRESH Message (5)
Address family identifier (AFI): IPv4 (1)
Subtype: Normal route refresh request [RFC2918] with/without ORF
[RFC5291] (0)
Subsequent address family identifier (SAFI): Unicast (1)
Attached below a tcpdump diff that perhaps can help improve visibility
:-)
Kind regards,
Job
Index: print-bgp.c
===================================================================
RCS file: /cvs/src/usr.sbin/tcpdump/print-bgp.c,v
retrieving revision 1.29
diff -u -p -r1.29 print-bgp.c
--- print-bgp.c 3 Jul 2019 03:24:03 -0000 1.29
+++ print-bgp.c 17 Jun 2021 13:24:33 -0000
@@ -97,7 +97,7 @@ struct bgp_route_refresh {
u_int16_t len;
u_int8_t type;
u_int8_t afi[2]; /* unaligned; should be u_int16_t */
- u_int8_t res;
+ u_int8_t subtype;
u_int8_t safi;
};
#define BGP_ROUTE_REFRESH_SIZE 23
@@ -189,6 +189,7 @@ static const char *bgp_capcode[] = {
/* 67: [Chen] */ "DYNAMIC_CAPABILITY",
/* 68: [Appanna] */ "MULTISESSION",
/* 69: [draft-ietf-idr-add-paths] */ "ADD-PATH",
+ /* 70: RFC7313 */ "ENHANCED_ROUTE_REFRESH"
};
#define bgp_capcode(x) \
@@ -199,7 +200,7 @@ static const char *bgpnotify_major[] = {
NULL, "Message Header Error",
"OPEN Message Error", "UPDATE Message Error",
"Hold Timer Expired", "Finite State Machine Error",
- "Cease", "Capability Message Error",
+ "Cease", "ROUTE_REFRESH Message Error",
};
#define bgp_notify_major(x) \
num_or_str(bgpnotify_major, \
@@ -323,6 +324,11 @@ static const char *afnumber[] = AFNUM_NA
num_or_str(afnumber, \
sizeof(afnumber)/sizeof(afnumber[0]), (x)))
+static const char *refreshtype[] = {
+ "Normal", "BoRR", "EoRR"
+};
+#define refresh_subtype(x) \
+ num_or_str(refreshtype, sizeof(refreshtype)/sizeof(refreshtype[0]), (x))
static const char *
num_or_str(const char **table, size_t siz, int value)
@@ -1069,7 +1075,8 @@ bgp_route_refresh_print(const u_char *da
bgp_route_refresh_header = (const struct bgp_route_refresh *)dat;
- printf(" (%s %s)",
+ printf(" %s (%s %s)",
+ refresh_subtype(bgp_route_refresh_header->subtype),
af_name(EXTRACT_16BITS(&bgp_route_refresh_header->afi)),
bgp_attr_nlri_safi(bgp_route_refresh_header->safi));
