David Lamparter [Tue, 24 Aug 2021 16:17:05 +0000 (18:17 +0200)]
pimd: fuse source_new() and add_source_by_addr()
This makes a lot more sense semantically (and matches the way groups are
handled.) Also allows placing additional restrictions on source
creation (e.g. limit on number of sources or ACLs.)
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
David Lamparter [Fri, 27 Aug 2021 08:45:16 +0000 (10:45 +0200)]
pimd: reformat after dropping indent levels
Intentionally separate here because the previous patch does a whole
bunch of "move stuff up 1 level of indentation", and reviewing that is
easier when you can use the ignore-whitespace option on diff.
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
David Lamparter [Tue, 24 Aug 2021 13:25:48 +0000 (15:25 +0200)]
pimd: IGMP memberships are not querier specific
IGMP group/source memberships are a property of the interface; the
particular IP address that the querier used to collect the data is
irrelevant.
... and IGMP packets get delivered only once to pimd anyway, since we
receive them on the "global" per-VRF IGMP socket. (The one in igmp_sock
is only used for sending queries.)
Signed-off-by: David Lamparter <equinox@opensourcerouting.org>
As such, using prefix2str() (or the %pFX format specifier) isn't
correct when logging such routes.
This commit adds a few special cases here and there to handle
OSPF6_DEST_TYPE_ROUTER routes differently. It'd probably be a good
idea to add a helper function to handle all cases in a single place,
but that can be left for a second moment.
Sai Gomathi [Wed, 18 Aug 2021 08:28:28 +0000 (01:28 -0700)]
pimd: IGMP groups are not getting timeout
Problem :
=======
When all the groups from Ixia are stopped,
groups still keep refreshing and not getting timeout
RCA:
====
IGMP Report is coming in include mode without any source address, this problem will come.
Fix :
===
If the requested filter mode is INCLUDE *and* the requested
source list is empty, then the entry corresponding to the
requested interface and multicast address is deleted if present.
If no such entry is present, the request is ignored.
When an interface receives the IGMP report without any source, then the group is deleted.
Signed-off-by: Sai Gomathi <nsaigomathi@vmware.com>
Renato Westphal [Thu, 19 Aug 2021 00:38:18 +0000 (21:38 -0300)]
ospf6d: remove incorrect debug guard
Also, update the ospf6_topo2 topotest since the expected output
was wrong. With this fix, NSSA routes will be created on r2
("redistribute connected"), and NSSA routes appear in the routing
table as regular external routes.
Renato Westphal [Wed, 18 Aug 2021 20:59:58 +0000 (17:59 -0300)]
ospf6d: fix flushing of all LSAs when NSSA is unconfigured
Once NSSA is unconfigured on an area, all self-originated Type-7
LSAs need to be flushed. The existing code was iterating over the
LSDB in the wrong way, causing ospf6_nssa_flush_area() to flush
LSAs of all types. Use the ALL_LSDB_TYPED_ADVRTR macro to perform
the intended iteration correctly.
Renato Westphal [Tue, 17 Aug 2021 00:35:27 +0000 (21:35 -0300)]
ospf6d: always unset the E-bit on NSSA ABRs
RFC 3101 - Section 2.1 says:
"(..) Interfaces associated with an NSSA will not send or receive
Type-5 LSAs on that interface but may send and receive Type-7 LSAs.
Therefore, if the N-bit is set in the options field, the E-bit must
be clear."
If the E-bit isn't cleared on an NSSA ABR, that will cause hello
packets to be dropped (due to parameters mismatch), which will
prevent the ABR from forming adjacencies with others routers in
the NSSA area.
This problem didn't affect the existing NSSA topotest by chance
of luck. In that topotest, in the NSSA ABR, the NSSA area is
configured before any interface is associated to it. That caused
ospf6_check_and_set_router_abr() to return false, leading to
the unsetting of the E-bit. With this fix, the order in which
areas/interfaces are configured shouldn't matter because the E-bit
will always be unset on NSSA areas.
Christian Hopps [Tue, 24 Aug 2021 01:40:38 +0000 (21:40 -0400)]
tests: use correct key for deleting entry.
OSPF mixes uses of "delete" and "del_action" depending on which library
function is called. It's a bug-prone mess that needs fixing; however, for
now we fix the one obvious incorrect use in this test.
Igor Ryzhov [Mon, 23 Aug 2021 20:54:12 +0000 (23:54 +0300)]
lib, zebra: move vrf netns commands from lib to zebra
"[no] netns NAME" commands are part of the lib, but they are actually
zebra-only:
- they are using vrf_netns_handler_create and its description clearly
says that it "should be called from zebra only"
- vtysh sends these commands only to zebra
- only zebra outputs the netns related config
- zebra notifies other daemons about netns attachment
Igor Ryzhov [Sun, 8 Aug 2021 19:38:50 +0000 (22:38 +0300)]
*: explicitly print "exit" at the end of every node config
There is a possibility that the same line can be matched as a command in
some node and its parent node. In this case, when reading the config,
this line is always executed as a command of the child node.
For example, with the following config:
```
router ospf
network 193.168.0.0/16 area 0
!
mpls ldp
discovery hello interval 111
!
```
Line `mpls ldp` is processed as command `mpls ldp-sync` inside the
`router ospf` node. This leads to a complete loss of `mpls ldp` node
configuration.
To eliminate this issue and all possible similar issues, let's print an
explicit "exit" at the end of every node config.
This commit also changes indentation for a couple of existing exit
commands so that all existing commands are on the same level as their
corresponding node-entering commands.
Igor Ryzhov [Thu, 5 Aug 2021 20:54:22 +0000 (23:54 +0300)]
pathd: rework config printing code
Instead of setting a config_write callback for each node, set a single
callback and print all config from there. It is necessary for the
following work on explicit "exit" command in every node.
Currently, in frr-reload we:
- store a list of single-line context keywords which needs to be
frequently updated,
- have a separate "if" clause for every node and subnode we have in FRR.
Instead, we can store the tree of all known FRR nodes. This tree needs
to be updated whenever we add a new node, which is not frequent. And,
most importantly, it allows us to write node-agnostic code and save more
than 250 LOC.
ospf6d: External LSAs reoriginates on every redistribute CLI
Problem Statement:
==================
Everytime redistribute CLI is executed, external LSAs are
re-originated. When there is no change in the CLI parameters
the LSAs should not get re-originated.
Fix:
=================
Check if the CLI params are same, do not re-originate the LSA.
Christian Hopps [Sat, 21 Aug 2021 07:04:42 +0000 (03:04 -0400)]
tests: uniq vrf names when using netns
- In order to run tests in parallel the netns-based vrfs need to
have unique names primarily bc they are all tracked/looked-up in
`/run/netns` which is not network namespace nesting friendly
- use ip(8) exclusively rather than a mix of `ip` and `ifconfig`
and `vconfig`, reducing required pkg count by a couple.
Christian Hopps [Fri, 20 Aug 2021 12:24:23 +0000 (08:24 -0400)]
tests: fix broken bgp GR test (non-deterministic)
- bugs in the support library function `verify_gr_address_family`
allowed this test to pass depending on ordering of python dictinoary
keys. Fix the bugs, fix the test.
Christian Hopps [Tue, 3 Aug 2021 22:22:09 +0000 (22:22 +0000)]
pimd: change ranges to 1..max, type align with IETF YANG.
While defaults are good picks for "reasonable" guesses, min and max
range values really aren't. Operators and experimenters often like to
configure "unreasonable" values to stress test, tests boundary
conditions and explore innovations.
With that in mind, change all ranges to 1..max (of type).
While we're here add optional ignored values in the "no" CLI forms.
Philippe Guibert [Tue, 17 Aug 2021 14:43:37 +0000 (16:43 +0200)]
bgpd: imported evpn rt5 routes copy igpmetric
when doing BGP over an IGP platform, the expectation is that
the path calculation for a given prefix takes into account the
igpmetric given by IGP.
This is true with prefixes obtained in a given BGP instance where
peering occurs. For instance, ipv4 unicast entries or l2vpn evpn
entries work this way. The igpmetric is obtained through nexthop
tracking, like below:
however, for imported EVPN RT5 entries, the igpmetric was not
copied from the parent path info. Fix it. In this way, the
imported route entries use the igpmetric of the parent pi.
Signed-off-by: Philippe Guibert <philippe.guibert@6wind.com>
ospfd: Summarised External LSA is not flushed in one scenario
Fix CI Failure test_ospf_type5_summary_tc45_p0
Problem Statement:
==================
Summarised LSA is not flushed in OSPFv2 in below scenario:
1. Configure summary-address in ospfv2
2. redistribute static and connected.
3. Check the LSAs are received on neighbor.
4. Now remove all OSPFv2 configs, so neighbor will still have the summarised LSA.
5. Configure router ospf with redistribute static and connected.
6. Check the DB, summarised LSA is present although the configuration is not present.
7. Now configure the summary-address and remove the configuration after sometime.
8. The summarised LSA will be still present.
RCA:
==================
When self originated LSA is received from the neighbor and that
LSA is summarised one, the LSA is refreshed but a flag is not set
due to which it was not able to remove it later.
Fix:
==================
Set the originated flag when refreshing summarised LSA.
pimd: IGMPv2 query timer gets delayed in a scenario
Problem Statement:
==================
IGMP query is sent at irregular intervals
(more than 30 seconds) when "ip igmp query-max-response-time 100"
command is executed multiple times.
RCA:
=================
When "ip igmp query-max-response-time 100" is executed, the timers
are reset resulting in the delay of sending the query.
Fix:
=================
When there is no change in the config value, we should not reset
the timers.
projects / mirror_frr.git / log