blob: 0a5617e4af6394ed4ba730ffa417c53055c512c4 [file] [log] [blame]
# $QuaggaId: $Format:%an, %ai, %h$ $
T1 DONE Implement debug command
test pim receive join
T2 DONE Implement debug command
test pim receive prune
T3 DONE Per-interface Downstream (S,G) state machine
(RFC 4601 4.5.3. Receiving (S,G) Join/Prune Messages)
T4 DONE Upstream (S,G) state machine
(RFC 4601 4.5.7. Sending (S,G) Join/Prune Messages)
T5 DONE Verify Data Packet Forwarding Rules
RFC 4601 4.2. Data Packet Forwarding Rules
RFC 4601 4.8.2. PIM-SSM-Only Routers
Additionally, the Packet forwarding rules of Section 4.2 can be
simplified in a PIM-SSM-only router:
iif is the incoming interface of the packet.
oiflist = NULL
if (iif == RPF_interface(S) AND UpstreamJPState(S,G) == Joined) {
oiflist = inherited_olist(S,G)
} else if (iif is in inherited_olist(S,G)) {
send Assert(S,G) on iif
}
oiflist = oiflist (-) iif
forward packet on all interfaces in oiflist
Macro:
inherited_olist(S,G) =
joins(S,G) (+) pim_include(S,G) (-) lost_assert(S,G)
T6 DONE Implement (S,G) Assert state machine (RFC 4601, section 4.6.1).
Changes in pim_ifchannel.ifassert_winner should trigger
pim_upstream_update_join_desired().
Depends on TODO T27.
Depends on TODO T33.
See also CAVEAT C7.
See also: RFC 4601 4.5.7. Sending (S,G) Join/Prune Messages
Transitions from Joined State
RPF'(S,G) changes due to an Assert
http://www.hep.ucl.ac.uk/~ytl/multi-cast/pim-dm_01.html:
The PIM Assert mechanism is used to shutoff duplicate flows onto
the same multiaccess network. Routers detect this condiction when
they receive an (S,G) packet via a multi-access interface that is
in the (S,G) OIL. This causes the routers to send Assert
Messages.
Note that neighbors will not accept Join/Prune or Assert messages
from a router unless they have first heard a Hello message from that
router. Thus, if a router needs to send a Join/Prune or Assert
message on an interface on which it has not yet sent a Hello message
with the currently configured IP address, then it MUST immediately
send the relevant Hello message without waiting for the Hello Timer
to expire, followed by the Join/Prune or Assert message.
T7 DONE Implement hello option: LAN Prune Delay
T8 DONE Implement J/P_Override_Interval(I)
Depends on TODO T7.
See pim_ifchannel.c, pim_ifchannel_prune(), jp_override_interval.
T9 DONE Detect change in IGMPv3 RPF interface/next-hop for S and update.
channel_oil vif index accordingly ?
Beware accidentaly adding looped MFC entries (IIF=OIF).
T10 DONE React to (S,G) join directed to another upstream address. See
also:
RFC 4601: 4.5.7. Sending (S,G) Join/Prune Messages
If a router wishes to propagate a Join(S,G) upstream, it must also
watch for messages on its upstream interface from other routers on
that subnet, and these may modify its behavior. If it sees a
Join(S,G) to the correct upstream neighbor, it should suppress its
own Join(S,G). If it sees a Prune(S,G), Prune(S,G,rpt), or
Prune(*,G) to the correct upstream neighbor towards S, it should
be prepared to override that prune by scheduling a Join(S,G) to be
sent almost immediately.
T11 DONE Review protocol modifications for SSM
(RFC 4601 4.8.1. Protocol Modifications for SSM Destination
Addresses)
T12 DONE Review updates of RPF entries.
FIXME pim_upstream.c send_join():
Currently only one upstream state is affected by detection of RPF change.
RPF change should affect all upstream states sharing the RPF cache.
T13 DONE Check that RFC macros using S,G,RPF_interface(S) are actually
implemented with this strategy:
rpf_ifch=find_ifch(up->rpf->interface).
See pim_rpf.c pim_rpf_find_rpf_addr() for a correct example.
$ grep -i macro pimd/*.c
pimd/pim_iface.c: RFC 4601: 4.1.6. State Summarization Macros
pimd/pim_ifchannel.c: RFC 4601: 4.6.5. Assert State Macros
pimd/pim_ifchannel.c: RFC 4601: 4.1.6. State Summarization Macros
pimd/pim_ifchannel.c: RFC 4601: 4.1.6. State Summarization Macros
pimd/pim_ifchannel.c: RFC 4601: 4.6.5. Assert State Macros
pimd/pim_ifchannel.c: Macro:
pimd/pim_rpf.c: RFC 4601: 4.1.6. State Summarization Macros
T14 DONE Send Assert(S,G) on iif as response to WRONGVIF kernel upcall.
See pim_mroute.c mroute_msg().
T15 DONE Interface command to statically join (S,G).
interface eth0
ip igmp join-group 239.1.1.1 source 1.1.1.1
T16 DONE RPF'(S,G) lookup is not working for S reachable with default route.
See "RPF'(S,G) not found" in pim_rpf_update() from pim_rpf.c.
Zebra daemon RIB is not reflecting changes in kernel routes
accurately?
T17 DONE Prevent CLI from creating bogus interfaces.
Example:
conf t
interface xxx
T18 Consider reliable pim solution (refresh reduction)
A Reliable Transport Mechanism for PIM
http://tools.ietf.org/wg/pim/draft-ietf-pim-port/
PORT=PIM-Over-Reliable-Transport
T19 DONE Fix self as neighbor
See mailing list post:
http://lists.gnu.org/archive/html/qpimd-users/2009-04/msg00000.html
T20 DONE Fix debug message: "pim_neighbor_update: internal error:
trying to replace same prefix list"
See mailing list post:
http://lists.gnu.org/archive/html/qpimd-users/2009-04/msg00000.html
T21 DONE Clean-up PIM/IGMP interface mismatch debugging
See option PIM_CHECK_RECV_IFINDEX_SANITY in pimd/Makefile.am
See mailing list post:
http://lists.nongnu.org/archive/html/qpimd-users/2009-04/msg00003.html
T22 DONE IGMP must be protected against adding looped MFC entries
created by both source and receiver attached to the same
interface.
T23 DONE libzebra crash after zclient_lookup_nexthop.
See mailing list post:
http://lists.nongnu.org/archive/html/qpimd-users/2009-04/msg00008.html
T24 DONE zserv may return recursive routes:
- nexthop type is set to ZEBRA_NEXTHOP_IPV4
- ifindex is not reported
- calls expecting ifindex (fib_lookup_if_vif_index) are disrupted
See also this mailing list post:
[PATCH 21/21] Link detect and recursive routes
http://www.gossamer-threads.com/lists/quagga/dev/17564
T25 DONE Zclient nexthop lookup missing OSPF route to 1.1.1.1/32
See also:
pim_zlookup.c zclient_lookup_nexthop misses OSPF 1.1.1.1/32
zebra/zebra_vty.c show_ip_route_addr_cmd hits OSPF 1.1.1.1/32
T26 DONE Zebra daemon is marking recursive static route as inactive.
FIXED: zebra daemon was incorrectly marking recursive routes
pointing to kernel routes as inactive:
zebra/zebra_rib.c nexthop_active_ipv4:
-- Original:
else if (CHECK_FLAG (rib->flags, ZEBRA_FLAG_INTERNAL))
-- Fixed:
else if (CHECK_FLAG (rib->flags, ZEBRA_FLAG_INTERNAL) ||
match->type == ZEBRA_ROUTE_KERNEL)
Old problem description:
This prevents rib_match_ipv4 from returning its nexthop:
client: pim_zlookup.c zclient_read_nexthop
server: zebra/zserv.c zsend_ipv4_nexthop_lookup_v2 -> rib_match_ipv4
Kernel route is injected into zebra in zebra_rib.c rib_add_ipv4
Examples:
rt_netlink.c:726: rib_add_ipv4 (ZEBRA_ROUTE_KERNEL, flags, &p, gate, src, index, table, metric, 0);
rt_netlink.c:864: rib_add_ipv4 (ZEBRA_ROUTE_KERNEL, 0, &p, gate, src, index, table, 0, 0);
This patch didn't fix the issue:
[PATCH 21/21] Link detect and recursive routes
http://www.gossamer-threads.com/lists/quagga/dev/17564
See the example below for the route 2.2.2.2.
bash# route add -host 1.1.1.1 gw 127.0.0.1
bash# route add -host 2.2.2.2 gw 1.1.1.1
bash# netstat -nvr
Kernel IP routing table
Destination Gateway Genmask Flags MSS Window irtt Iface
2.2.2.2 1.1.1.1 255.255.255.255 UGH 0 0 0 lo
1.1.1.1 127.0.0.1 255.255.255.255 UGH 0 0 0 lo
192.168.0.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0
0.0.0.0 192.168.0.2 0.0.0.0 UG 0 0 0 eth0
bash#
zebra# sh ip route
Codes: K - kernel route, C - connected, S - static, R - RIP, O - OSPF,
I - ISIS, B - BGP, > - selected route, * - FIB route
K>* 0.0.0.0/0 via 192.168.0.2, eth0
K>* 1.1.1.1/32 via 127.0.0.1, lo
K * 2.2.2.2/32 via 1.1.1.1, lo inactive
C>* 127.0.0.0/8 is directly connected, lo
C>* 192.168.0.0/24 is directly connected, eth0
quagga-pimd-router# sh ip route 1.1.1.1
Address NextHop Interface Metric Preference
1.1.1.1 127.0.0.1 lo 0 0
quagga-pimd-router#
quagga-pimd-router# sh ip route 2.2.2.2
Address NextHop Interface Metric Preference
2.2.2.2 192.168.0.2 eth0 0 0
quagga-pimd-router#
T27 DONE Implement debug command
test pim receive assert
See also TODO T6: (S,G) Assert state machine.
T28 DONE Bad IPv4 address family=02 in Join/Prune dump
Reported by Andrew Lunn <andrew.lunn@ascom.ch>
# 58-byte pim v2 Join/Prune dump
# ------------------------------
# IPv4 address family=02 is wrong, correct IPv4 address family is 01
# See http://www.iana.org/assignments/address-family-numbers
#
c8XX YY03 : ip src 200.xx.yy.3
e000 000d : ip dst 224.0.0.13
9404 0000 : ip router alert option 148.4.0.0
2300 ab13 : pimv2,type=3 res=00 checksum=ab13
0200 : upstream family=02, encoding=00
c8XX YY08 : upstream 200.xx.yy.8
0001 00d2 : res=00 groups=01 holdtime=00d2
0200 0020 : group family=02, encoding=00, res=00, mask_len=20
ef01 0101 : group address 239.1.1.1
0001 0000 : joined=0001 pruned=0000
0200 0020 : source family=02, encoding=00, res=00, mask_len=20
0101 0101 : source address 1.1.1.1
T29 DONE Reset interface PIM-hello-sent counter when primary address changes
See pim_ifp->pim_ifstat_hello_sent
RFC 4601: 4.3.1. Sending Hello Messages
Thus, if a router needs to send a Join/Prune or Assert message on
an interface on which it has not yet sent a Hello message with the
currently configured IP address, then it MUST immediately send the
relevant Hello message without waiting for the Hello Timer to
expire, followed by the Join/Prune or Assert message.
T30 DONE Run interface DR election when primary address changes
Reported by Andrew Lunn <andrew.lunn@ascom.ch>
See pim_if_dr_election().
T31 If an interface changes one of its secondary IP addresses, a Hello
message with an updated Address_List option and a non-zero
HoldTime should be sent immediately.
See also CAVEAT C15.
See also RFC 4601: 4.3.1. Sending Hello Messages
T32 Detection of interface primary address changes may fail when there
are multiple addresses.
See also CAVEAT C14.
pim_find_primary_addr() should return interface primary address
from connected list. Currently it returns the first address.
Zebra daemon "show int" is able to keep the primary address as
first address.
T33 DONE Implement debug command: test pim receive upcall
See also TODO T6: (S,G) Assert state machine.
T34 DONE assert_action_a1
T35 DONE Review macros depending on interface I.
See also: grep ,I\) pimd/*.c
For the case (S,G,I) check if I is either
1) interface attached to this per-interface S,G state (don't think so)
or
2) an arbitrary interface (most probably)
For the arbitrary interface case (2), consider representing
interface ifp as its primary address (struct in_addr ifaddr). The
benefit is in_addr does not need to be dereferenced, so it does
not demand protection against crashes.
T36 DONE React to zebra daemon link-detect up/down notification.
pim_ifp->primary_address is managed by detect_primary_address_change()
depending on to ifp->connected (managed by zebra_interface_address_read()).
T37 DONE Review list of variables which may affect pim_upstream.c
pim_upstream_evaluate_join_desired().
Call pim_upstream_update_join_desired() accordingly.
See the order of invokation:
pim_if_dr_election(ifp);
pim_if_update_join_desired(pim_ifp); /* depends on DR */
pim_if_update_could_assert(ifp); /* depends on DR */
pim_if_update_my_assert_metric(ifp); /* depends on could_assert */
join_desired depends on:
pim_ifp->primary_address
pim_ifp->pim_dr_addr
ch->ifassert_winner_metric
ch->ifassert_winner
ch->local_ifmembership
ch->ifjoin_state
ch->upstream->rpf.source_nexthop.mrib_metric_preference
ch->upstream->rpf.source_nexthop.mrib_route_metric
ch->upstream->rpf.source_nexthop.interface
T38 DONE Detect change in AssertTrackingDesired(S,G,I)
See the order of invokation:
dr_election: none
update_join_desired: depends on DR
update_tracking_desired: depends on DR, join_desired
AssertTrackingDesired(S,G,I) depends on:
pim_ifp->primary_address
pim_ifp->pim_dr_addr
ch->local_ifmembership
ch->ifassert_winner
ch->ifjoin_state
ch->upstream->rpf.source_nexthop.interface
PIM_UPSTREAM_FLAG_TEST_DR_JOIN_DESIRED(ch->upstream->flags)
T39 DONE AssertTrackingDesired: flags is not matching evaluation
# show ip pim assert-internal
CA: CouldAssert
ECA: Evaluate CouldAssert
ATD: AssertTrackingDesired
eATD: Evaluate AssertTrackingDesired
Interface Address Source Group CA eCA ATD eATD
eth0 192.168.1.100 1.1.1.1 239.1.1.1 no no no yes
#
T40 Lightweight MLDv2
http://tools.ietf.org/html/draft-ietf-mboned-lightweight-igmpv3-mldv2-05
http://www.ietf.org/internet-drafts/draft-ietf-mboned-lightweight-igmpv3-mldv2-05.txt
http://www.ietf.org/html.charters/mboned-charter.html
T41 DONE ssmping support
See also:
http://www.venaas.no/multicast/ssmping/
draft-ietf-mboned-ssmping-07
http://tools.ietf.org/html/draft-ietf-mboned-ssmping-07
Example:
debug ssmpingd
conf t
ip ssmpingd 1.1.1.1
show ip ssmpingd
T42 Static igmp join fails when loading config at boot time
! Wrong behavior seen at boot time:
!
2010/02/22 08:59:00 PIM: igmp_source_forward_start: ignoring request for
looped MFC entry (S,G)=(3.3.3.3,239.3.3.3): igmp_sock=12 oif=eth0 vif_index=2
! Correct behavior seen later:
!
2010/02/22 09:03:16 PIM: igmp_source_forward_start: ignoring request for
looped MFC entry (S,G)=(2.2.2.2,239.2.2.2): igmp_sock=17 oif=lo vif_index=1
! To see the wrong message at boot:
!
debug igmp trace
!
interface lo
ip igmp
ip igmp join 239.2.2.2 2.2.2.2
ip igmp join 239.3.3.3 3.3.3.3
!
! Interfaces indexes:
Interface Address ifi Vif PktsIn PktsOut BytesIn BytesOut
eth0 200.202.112.3 2 2 0 0 0 0
lo 127.0.0.1 1 1 0 0 0 0
-x-