Revision 18c8241a TODO
| TODO | ||
|---|---|---|
| 1 | 1 |
Core |
| 2 | 2 |
~~~~ |
| 3 |
- route validation |
|
| 4 | 3 |
- fake multipath? |
| 5 | 4 |
- config file: symbolic constants? |
| 6 | 5 |
- counters (according to SNMP MIB?) |
| 7 | 6 |
- generation of subnet mask ICMP's for v6? |
| 8 |
- debugging dumps and protocol tracing! |
|
| 9 | 7 |
- unaligned accesses? |
| 8 |
- neighbor cache: local broadcast address? |
|
| 9 |
- ipv4: recognize site scope addresses? |
|
| 10 |
- ifdef out some debugging code? |
|
| 11 |
- better memory allocators |
|
| 12 |
- precedence of all packets (incl. TCP) |
|
| 13 |
- default preferences of protocols: prefer BGP over OSPF/RIP external routes? |
|
| 14 |
- all internal tables are in host order |
|
| 10 | 15 |
|
| 16 |
- filter: logging of dropped routes (?) |
|
| 17 |
- limitation of memory consumption: per-process and total (?) |
|
| 18 |
- alloca |
|
| 19 |
- adding of route: clear all bits not covered by masklen |
|
| 20 |
- switch: generate default route only if at least one BGP connection exists |
|
| 11 | 21 |
|
| 12 |
RIP |
|
| 13 |
~~~ |
|
| 14 |
- RIP: export-only and import-only mode? |
|
| 15 |
- drop RIPv1 (Historic protocol)? |
|
| 16 |
|
|
| 17 |
OSPF |
|
| 18 |
~~~~ |
|
| 19 |
|
|
| 20 |
Almquist & Kastenholz [Page 111] |
|
| 21 |
RFC 1716 Towards Requirements for IP Routers November 1994 |
|
| 22 |
- route recalculation timing + flap dampening |
|
| 22 | 23 |
|
| 24 |
- reconfiguration without restart of all protocols? |
|
| 25 |
- change of interface address: ??? (down and up?) |
|
| 26 |
- "generate default route" switch for all IGP's |
|
| 23 | 27 |
|
| 24 |
7.2.2.2 Specific Issues |
|
| 28 |
- running protocol on an interface: |
|
| 29 |
- interface is not required to exist |
|
| 30 |
- can specify a wildcard pattern or an interface list |
|
| 25 | 31 |
|
| 26 |
Virtual Links |
|
| 32 |
- timers - one-shot and periodic, resolution 1 sec, randomized |
|
| 33 |
- re-configuration: restart of routing protocols (shutdown mode) |
|
| 34 |
- route: originating AS |
|
| 27 | 35 |
|
| 28 |
There is a minor error in the specification that can cause |
|
| 29 |
routing loops when all of the following conditions are |
|
| 30 |
simultaneously true: |
|
| 36 |
- Check incoming packets and log errors!! |
|
| 31 | 37 |
|
| 32 |
(1) A virtual link is configured through a transit area, |
|
| 33 | 38 |
|
| 34 |
(2) Two separate paths exist, each having the same |
|
| 35 |
endpoints, but one utilizing only non-virtual |
|
| 36 |
backbone links, and the other using links in the |
|
| 37 |
transit area, and |
|
| 39 |
RIP |
|
| 40 |
~~~ |
|
| 41 |
- RIP: export-only and import-only mode? |
|
| 42 |
- drop RIPv1 (Historic protocol)? |
|
| 43 |
- Route Tag |
|
| 44 |
- limit routing table xfer (frequency, only to neighbors) |
|
| 45 |
- multicast on/off |
|
| 46 |
- remember routes for all neighbors? |
|
| 38 | 47 |
|
| 39 |
(3) The latter path is part of the (underlying physical |
|
| 40 |
representation of the) configured virtual link, |
|
| 41 |
routing loops may occur. |
|
| 48 |
OSPF |
|
| 49 |
~~~~ |
|
| 50 |
- Dijkstra: use Fibonacci heaps? |
|
| 51 |
- point-to-point interface with address: advertise as stub network |
|
| 52 |
- static routes: stub networks? |
|
| 53 |
- modes: PtP, PtP-unnumbered, Broadcast, NBMA, point-to-multipoint |
|
| 54 |
- importing of device routes for networks where we don't run OSPF |
|
| 55 |
- tie breaking for equal type 2 ext metrics by using internal (type 1) metric |
|
| 56 |
- SPF tree recalc timing (per-area timers?) |
|
| 57 |
- aggregation: specify network list for each area |
|
| 58 |
- stub area: either no external routes or only default route |
|
| 59 |
- automatic generation of external route tags (RFC1403) |
|
| 42 | 60 |
|
| 43 |
To prevent this, an implementation of OSPF SHOULD invoke |
|
| 44 |
the calculation in Section 16.3 of [ROUTE:1] whenever any |
|
| 45 |
part of the path to the destination is a virtual link (the |
|
| 46 |
specification only says this is necessary when the first |
|
| 47 |
hop is a virtual link). |
|
| 48 | 61 |
|
| 49 | 62 |
BGP |
| 50 | 63 |
~~~ |
| 51 |
- BGP: |
|
| 52 | 64 |
- in, local, out RIB |
| 53 | 65 |
- maxsize=4096 |
| 54 | 66 |
- BGP identifier aka router id |
| 55 |
- removal of loops
|
|
| 67 |
- detection of loops
|
|
| 56 | 68 |
- aggregation, ATOMIC_AGGREGATE |
| 57 | 69 |
- communities |
| 58 | 70 |
- confederations |
| ... | ... | |
| 71 | 83 |
- expected neighbor AS |
| 72 | 84 |
- hold time |
| 73 | 85 |
- idle timer after error: initial value, exponential growth, maximum value |
| 74 |
|
|
| 75 |
- address testing macros (ALL_ZEROS) |
|
| 76 |
- all internal tables are in network order (?) |
|
| 77 |
- logging of errors and debug dumps |
|
| 78 |
- filter: logging of dropped routes (?) |
|
| 79 |
- limitation of memory consumption: per-process and total |
|
| 80 |
- alloca |
|
| 81 |
- precedence of all packets (incl. TCP) |
|
| 82 |
- adding of route: clear all bits not covered by masklen |
|
| 83 |
- switch: generate default route only if at least one BGP connection exists |
|
| 84 |
|
|
| 85 |
- route update: new, change, remove |
|
| 86 |
- route recalculation timing |
|
| 87 |
|
|
| 88 |
- CONFIG_TOS |
|
| 89 |
- CONFIG_MULTIPATH |
|
| 90 |
|
|
| 91 |
- reconfiguration without restart of all protocols? |
|
| 92 |
- change of interface address: ??? (down and up?) |
|
| 93 |
- "generate default route" switch for all IGP's |
|
| 94 |
|
|
| 95 |
- RIPv2: |
|
| 96 |
- Route Tag |
|
| 97 |
- limit routing table xfer (frequency, only to neighbors) |
|
| 98 |
- multicast on/off |
|
| 99 |
- remember routes for all neighbors? |
|
| 100 |
|
|
| 101 |
- BGP: |
|
| 102 | 86 |
- import of IGP routes (use external route tags from OSPF) |
| 103 |
|
|
| 104 |
- Interface: |
|
| 105 |
- RIP metric |
|
| 106 |
- multicast capability flag |
|
| 107 |
- MTU |
|
| 108 |
- OSPF metrics (per-TOS) |
|
| 109 |
|
|
| 110 |
- running protocol on an interface: |
|
| 111 |
- interface is not required to exist |
|
| 112 |
- can specify a wildcard pattern or an interface list |
|
| 113 |
|
|
| 114 |
- preferences: |
|
| 115 |
- directly connected |
|
| 116 |
- static |
|
| 117 |
- OSPF internal, OSPF ext type 1 (comparable metrics), OSPF inter-area |
|
| 118 |
- RIP |
|
| 119 |
- BGP |
|
| 120 |
- OSPF ext type 2 |
|
| 121 |
- sink |
|
| 122 |
|
|
| 123 |
- lib: |
|
| 124 |
- MD5 |
|
| 125 |
|
|
| 126 |
- OSPF: |
|
| 127 |
- Dijkstra: use Fibonacci heaps? |
|
| 128 |
- point-to-point interface with address: advertise as stub network |
|
| 129 |
- static routes: stub networks? |
|
| 130 |
- modes: PtP, PtP-unnumbered, Broadcast, NBMA, point-to-multipoint |
|
| 131 |
- importing of device routes for networks where we don't run OSPF |
|
| 132 |
- tie breaking for equal type 2 ext metrics by using internal (type 1) metric |
|
| 133 |
- SPF tree recalc timing (per-area timers?) |
|
| 134 |
- aggregation: specify network list for each area |
|
| 135 |
- stub area: either no external routes or only default route |
|
| 136 |
- automatic generation of external route tags (RFC1403) -- what about |
|
| 137 |
using the same rule for RIPv2? [shared code?] |
|
| 138 |
|
|
| 139 |
- timers - one-shot and periodic, resolution 1 sec, randomized |
|
| 140 |
- re-configuration: restart of routing protocols (shutdown mode) |
|
| 141 |
- route: originating AS |
|
| 142 |
|
|
| 143 |
- Check incoming packets and log errors!! |
|
Also available in: Unified diff