MPLS VPN networks can either peer with iBGP or eBGP. When
calculating the distance to send to zebra, the imported prefix
is never sent with distance information, even if the vty
command is used under the ipv4 unicast address family:
router bgp 65505 vrf vrf1
address-family ipv4 unicast
distance bgp 26 27 28
[vpn config]
The observation is that the distance sent to zebra for an
imported prefix is still 20:
[..]
VRF vrf1:
B> 192.168.0.0/24 [20/0] via 2.2.2.2 (vrf default) (recursive), label 20, weight 1, 00:00:12
* via 10.125.0.6, ntfp3 (vrf default), label implicit-null/20, weight 1, 00:00:12
The expectation is that the incoming prefix has to follow the
distance that is configured, or the distance derived from the peer
relationship established by the parent prefix.
In the case, an iBGP relationship is done, and no distance
configuration is done, the below show is expected:
[..]
VRF vrf1:
B*> 192.168.0.0/24 [200/0] via 192.168.0.2, r1-gre0 (vrf default), label 20, weight 1, 00:00:12
In the case an iBGP relationship is done, and distance configuration
is performed as below:
[..]
distance bgp 21 201 41
[..]
Then the below show is expected:
[..]
VRF vrf1:
B*> 192.168.0.0/24 [201/0] via 192.168.0.2, r1-gre0 (vrf default), label 20, weight 1, 00:00:12
To get this behaviour, get the peer origin where the prefix is coming
from.
Signed-off-by: Philippe Guibert <philippe.guibert@6wind.com>
[New LWP 2524]
[New LWP 2539]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Core was generated by `/opt/avi/bin/bgpd -f /run/frr/avi_ns3_bgpd.config -i /opt/avi/etc/avi_ns3_bgpd.'.
Program terminated with signal SIGABRT, Aborted.
[Current thread is 1 (Thread 0x7f92ac8f1740 (LWP 2524))]
0 0x00007f92acb3800b in raise () from /lib/x86_64-linux-gnu/libc.so.6
[Current thread is 1 (Thread 0x7f92ac8f1740 (LWP 2524))]
0 0x00007f92acb3800b in raise () from /lib/x86_64-linux-gnu/libc.so.6
1 0x00007f92acb17859 in abort () from /lib/x86_64-linux-gnu/libc.so.6
2 0x00007f92acb17729 in ?? () from /lib/x86_64-linux-gnu/libc.so.6
3 0x00007f92acb28fd6 in __assert_fail () from /lib/x86_64-linux-gnu/libc.so.6
4 0x00007f92accf2164 in pthread_mutex_lock () from /lib/x86_64-linux-gnu/libpthread.so.0
5 0x000055b46be1ef63 in bgp_keepalives_wake () at bgpd/bgp_keepalives.c:311
6 0x000055b46be1f111 in bgp_keepalives_stop (fpt=0x55b46cfacf20, result=<optimized out>) at bgpd/bgp_keepalives.c:323
7 0x00007f92acea9521 in frr_pthread_stop (fpt=0x55b46cfacf20, result=result@entry=0x0) at lib/frr_pthread.c:176
8 0x00007f92acea9586 in frr_pthread_stop_all () at lib/frr_pthread.c:188
9 0x000055b46bdde54a in bgp_pthreads_finish () at bgpd/bgpd.c:8150
10 0x000055b46bd696ca in bgp_exit (status=0) at bgpd/bgp_main.c:210
11 sigint () at bgpd/bgp_main.c:154
12 0x00007f92acecc1e9 in quagga_sigevent_process () at lib/sigevent.c:105
13 0x00007f92aced689a in thread_fetch (m=m@entry=0x55b46cf23540, fetch=fetch@entry=0x7fff95379238) at lib/thread.c:1487
14 0x00007f92aceb2681 in frr_run (master=0x55b46cf23540) at lib/libfrr.c:1010
15 0x000055b46bd676f4 in main (argc=11, argv=0x7fff953795a8) at bgpd/bgp_main.c:482
Root cause:
This is due to race condition between main thread & keepalive thread during clean-up.
This happens when the keepalive thread is processing a wake signal owning the mutex, when meanwhile the main thread tries to stop the keepalives thread.
In main thread, the keepalive thread’s running bit (fpt->running) is set to false, without taking the mutex & then it blocks on mutex.
Meanwhile, keepalive thread which owns the mutex sees that the running bit is false & executes bgp_keepalives_finish() which also frees up mutex.
Main thread that is waiting on mutex with pthread_mutex_lock() will cause core while trying to access mutex.
Fix:
Take the lock in main thread while setting the fpt->running to false.
Signed-off-by: Samanvitha B Bhargav <bsamanvitha@vmware.com>
Before this patch we allowed importing routes between VRFs in the same node,
only for external routes, but not for local (e.g.: redistribute).
Relax here a bit, and allow importing local routes between VRFs when the RT
list is modified using route reflectors.
Signed-off-by: Donatas Abraitis <donatas@opensourcerouting.org>
Before this patch, we needed to explicitly define a neighbor to be SOLO
(= separate update-group). Let's ease this functionality for an operator to
avoid confusions.
Signed-off-by: Donatas Abraitis <donatas@opensourcerouting.org>
The idea is to drop unwanted attributes from the BGP UPDATE messages and
continue by just ignoring them. This improves the security, flexiblity, etc.
This is the command that Cisco has also.
Signed-off-by: Donatas Abraitis <donatas@opensourcerouting.org>
For now, if the order was mixed, most of the commands are just silently
ignored. Let the operator notice that.
Signed-off-by: Donatas Abraitis <donatas@opensourcerouting.org>
Introduce a "detail" keyword for per-neighbor/per-afi-safi
advertised-routes and received-routes show commands.
Includes json support.
Signed-off-by: Trey Aspelund <taspelund@nvidia.com>
Added ipv4 and ipv6 option to existing "show bgp nexthop"
command to be able to query nexthops that belong to a
particular address-family.
Also fixed the warnings of MR 12171
Signed-off-by: Pooja Jagadeesh Doijode <pdoijode@nvidia.com>
When creating one interface "vxlan66" ( ip link add vxlan66 type vxlan ... ),
which initially maintains down status, saw one unexpected EC log:
```
zebra[37906]: [ZAG0W-VSNSD] interface vxlan66 vrf default(0) index 35 becomes active.
zebra[37906]: [HPWGA-Y527W] IFLA_VXLAN_LINK missing from VXLAN IF message
...
zebra[37906]: [W6BZR-YZPAB] RTM_NEWLINK update for vxlan66(35) sl_type 0 master 0 flags 0x1002
zebra[37906]: [MR3ZF-ATDBY] Intf vxlan66(35) has gone DOWN
zebra[37906]: [T44X9-FFNVB] Intf vxlan66(35) L2-VNI 66 is DOWN
zebra[37906]: [KEGYY-K8XVV] Send EVPN_DEL 66 to bgp
zebra[37906]: [HPWGA-Y527W] IFLA_VXLAN_LINK missing from VXLAN IF message
bgpd[37911]: [TV0XP-3WR0A] Rx VNI del VRF default VNI 66 tenant-vrf default SVI ifindex 0
bgpd[37911]: [MDW89-YAXJG][EC 33554497] 0: VNI hash entry for VNI 66 not found at DEL
```
Since commit `6f908ded80eeba40a850e8d1f6246fb3ed31e648` support interfaces
from down to down, and bgpd doesn't know "VNI 66" at all. So, remove this
EC log.
Signed-off-by: anlan_cs <vic.lan@pica8.com>
On shutdown, bgp calls an unlock for the bnc connected table,
via the bgp_connected_cleanup function. This function is
only ever called on shutdown, so we know that bgp is going
away. The refcount for the connected data can be more than
1. Let's not worry about the refcount on shutdown and
just delete the nodes instead of leaving them around.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
Shutdown of bgp results in both the bgp_path_info,
bgp_dest and bgp_table's not being freed because
the bgp_path_info remains locked.
Effectively static routes are scheduled for deletion but bgp_process
skips the work because the work queue sees that the bgp router
is marked for deletion. Effectively not doing any work and leaving
data on the floor.
Modify the code when attempting to put into the work queue to
notice and not do so but just unlock the path info.
This is effectively the same as what goes on for normal peering
as that it checks for shutdown and just calls bgp_path_info_free
too.
Signed-off-by: Donald Sharp <sharpd@nvidia.com>
Take it into consideration for one debug log:
EVPN MAC-IP routes with a L3 NHG id, has no nexthops.
Not "delete", but "add".
Before:
```
Tx route delete VRF 21 192.168.30.253/32 metric 0 tag 0 count 0 nhg 72580649
```
After:
```
Tx route add VRF 21 192.168.30.253/32 metric 0 tag 0 count 0 nhg 72580649
```
Signed-off-by: anlan_cs <vic.lan@pica8.com>
Use `DEFPY` to provide the `struct prefix` required by the debug
insertion functions. While here exchange `int` with `bool` where
appropriated.
Signed-off-by: Rafael Zalamena <rzalamena@opensourcerouting.org>
