Joanne Mikkelson [Wed, 17 Mar 2021 22:30:14 +0000 (15:30 -0700)]
bgpd: Allow next hop recursion in zebra if any multipath is not eBGP
When "bgp bestpath peer-type multipath-relax" is enabled, multipaths
with both eBGP and iBGP learned routes may exist. It is not desirable
for the iBGP next hops to be discarded from the FIB because they are not
directly connected. When publishing a nexthop group to zebra, the
ZEBRA_FLAG_ALLOW_RECURSION flag is normally not set when the best path
is eBGP; when "bgp bestpath aspath multipath-relax" is configured, the
flag will now be set if any paths are from iBGP peers. This leaves
all-eBGP multipaths still requiring nexthops over connected routes.
This new BGP configuration is akin to "bgp bestpath aspath
multipath-relax". When applied, paths learned from different peer types
will be eligible to be considered for multipath (ECMP). Paths from all
of eBGP, iBGP, and confederation peers may be included in multipaths
if they are otherwise equal cost.
This change preserves the existing bestpath behavior of step 10's result
being returned, not the result from steps 8 and 9, in the case where
both 8+9 and 10 determine a winner.
Igor Ryzhov [Tue, 16 Mar 2021 19:09:27 +0000 (22:09 +0300)]
ospfd: fix "show ip ospf database" issues
Current implementation of commands `show_ip_ospf_instance_database_cmd`
and `show_ip_ospf_instance_database_type_adv_router_cmd` have the
following problems:
- they doesn't have "vrf all" argument, however the processing of this
argument is implemented,
- they incorrectly implement json output for instances - they don't
output anything to the vty and don't release the json object.
To fix the problems, let's do the following:
1. Split `show_ip_ospf_instance_database_cmd` into two aliases to
`show_ip_ospf_database_max_cmd` and `show_ip_ospf_instance_database_max_cmd`.
The code is the same and doesn't need to be duplicated.
2. Split `show_ip_ospf_instance_database_type_adv_router_cmd` into two
separate functions - one regular and one for instances, which now
correctly implements the processing for json output.
vivek [Sun, 21 Mar 2021 20:39:51 +0000 (13:39 -0700)]
bgpd: Reset LLA NHT's interface if there is a change
For link-local IPv6 next hops, the next hop tracking is implemented based
on interface status changes. For this purpose, the ifindex is stored in
the NHT. Reset this value if a change in ifindex is noticed, such as for
example after a restart of the networking service.
Also add some additional debug logs.
Signed-off-by: Vivek Venkatraman <vivek@nvidia.com>
Updates: "bgpd: Switch LL nexthop tracking to be interface based"
Ticket: RM 2575386
Testing Done:
1. Manual verification
2. Precommit (#156), evpn-smoke (#155), bgp-smoke (#157), vrl (#158)
-- Precommit is clean, reported failures in evpn-smoke & vrl are resolved
-- some other tests fail in evpn-smoke, bgp-smoke & vrl, appear to be existing
-- or unrelated failures
David Lamparter [Tue, 2 Mar 2021 19:06:24 +0000 (20:06 +0100)]
lib: get rid of flog() usage in NB code
This parametrized use of flog with variable EC and priority doesn't mesh
particularly well with the xref code & there isn't really much reason to
not use fixed/constant calls like this.
Signed-off-by: David Lamparter <equinox@diac24.net>
no point in scheduling an LSP refresh immediately if we know it is
going to be postponed again due to the network still being in its
instability grace period
Signed-off-by: Emanuele Di Pascale <emanuele@voltanet.io>
when we receive an event from BFDD and we end up throwing it away,
make sure that we log (with debug guards) the reason for this, so
we can troubleshoot issues like the one addressed by the previous
commit.
Signed-off-by: Emanuele Di Pascale <emanuele@voltanet.io>
A wrong check was silently skipping the initialization of the bfd_session
struct in the adjacency if the router was not configured for IPv6. This
would cause BFD events to be ignored regardless of the configuration.
Also add a function to return the "name" of an adjacency and use it in a
couple of places, including the new log, instead of repeating the same
code in a bunch of places.
Signed-off-by: Emanuele Di Pascale <emanuele@voltanet.io>
Chirag Shah [Thu, 18 Mar 2021 05:04:33 +0000 (22:04 -0700)]
doc: add seqno to bgp as path list section
Add sequence no option to bgp as-path list cli syntax.
Add sequence no to example config.
Add auto generated sequence no in running-config if its not
provided in config.
David Lamparter [Tue, 16 Mar 2021 05:11:15 +0000 (06:11 +0100)]
tests: simplify docker rsync & support worktrees
The previous, more complex mechanism failed to take into account that
git worktrees only have a stub .git file & copying the worktree itself
is not enough. Just extract a file list beforehand & don't bother with
git inside the container.
Signed-off-by: David Lamparter <equinox@diac24.net>
David Lamparter [Wed, 17 Mar 2021 05:39:34 +0000 (06:39 +0100)]
pathd: remove mid-string line breaks
cf. workflow.rst ("lines over 80 characters are allowed for text strings
to make it possible to search the code for them"), matching Linux kernel
coding style.
Signed-off-by: David Lamparter <equinox@diac24.net>
David Lamparter [Tue, 2 Mar 2021 18:40:07 +0000 (19:40 +0100)]
lib: temporary workaround for LabN CI
Accept macros without ; for LabN CI *only*. This is a bit hairy since
we can't generate warnings for this, so it's very limited in both scope
and duration.
Signed-off-by: David Lamparter <equinox@diac24.net>
David Lamparter [Sun, 21 Feb 2021 05:18:10 +0000 (06:18 +0100)]
*: require semicolon after DEFINE_MTYPE & co
Back when I put this together in 2015, ISO C11 was still reasonably new
and we couldn't require it just yet. Without ISO C11, there is no
"good" way (only bad hacks) to require a semicolon after a macro that
ends with a function definition. And if you added one anyway, you'd get
"spurious semicolon" warnings on some compilers...
With C11, `_Static_assert()` at the end of a macro will make it so that
the semicolon is properly required, consumed, and not warned about.
Consistently requiring semicolons after "file-level" macros matches
Linux kernel coding style and helps some editors against mis-syntax'ing
these macros.
Signed-off-by: David Lamparter <equinox@diac24.net>
Gaurav Goyal [Thu, 11 Mar 2021 00:46:37 +0000 (13:46 +1300)]
nhrpd: Retry IPSec if NHRP is repeatedly failing
This prevents a failed IPSec connection from preventing DMVPN from working.
A failure situation can be reproduced using a Cisco peer, and and disabling then
re-enabling the tunnel IPSec protection (after the IPSec connection has
already been established).
Donald Sharp [Wed, 17 Mar 2021 02:28:29 +0000 (22:28 -0400)]
bgpd: If we have a SAFI conflict do not allow labeled unicast to reset
If we have a SAFI conflict, ie we are trying to activate safi's
UNICAST and LABELED_UNICAST at the same time, we should not
cause bestpath to be rerun and we should not try to put
labels on everything.
ckishimo [Tue, 16 Mar 2021 23:19:50 +0000 (00:19 +0100)]
tests: fix warning when checking ospfv3 convergence
The following error is shown when running the OSPFv3 tests
2021-03-16 23:37:44,792 INFO: Function returned global name 'data_rid' is not defined
2021-03-16 23:37:44,792 INFO: Retry [#1] after sleeping for 2s
2021-03-16 23:37:46,794 INFO: Verifying OSPF6 neighborship on router r1:
2021-03-16 23:37:46,993 INFO: Output for command [ show ipv6 ospf6 neighbor ] on router r1:
Neighbor ID Pri DeadTime State/IfState Duration I/F[State]
2.2.2.2 1 00:00:03 Full/PointToPoint 00:00:01 r1-r2-eth0[PointToPoint]
Fix the "data_rid" warning by using the correct variable
ckishimo [Tue, 16 Mar 2021 22:47:18 +0000 (23:47 +0100)]
ospf6d: fix iface commands lost when removing from area
In OSPFv3 when removing the interface from an area, all ospf6
interface commands are lost, so when changing the area you need
to reconfigure all ospf6 interface commands again
r1# conf t
r1(config)# router ospf6
r1(config-ospf6)# no interface r1-r2-eth0 area 0.0.0.0
r1(config-ospf6)# exit
r1# sh run
interface r1-r2-eth0
ipv6 address 2013:12::1/64
! <----- missing all ipv6 ospf6 commands
router ospf6
ospf6 router-id 1.1.1.1
!
This is because the interface is being deleted instead of disabled
(see PR#7717) I believe the interface should be left as disabled
(not deleted) when removing the interface from the area
projects / mirror / frr.git / log