Micro-bfd sessions with timer “10 ms x 5” flaps after a route churn or LSP flap event.
FPC5(labdevice-RE0 vty)# show ppm transmits
Protocol Tx Interval PPM handle Inline Quick Xmit
(msec)
BFD 10 1525 No Yes
BFD 10 1519 No Yes
BFD 10 1527 No Yes
BFD 10 1526 No Yes
BFD 10 1522 No Yes
BFD 10 1520 No Yes
BFD 10 1523 No Yes
BFD 10 1518 No Yes
May 1 19:48:01 2025 labdevice-RE1 rpd[33264]: RPD_MPLS_PATH_DOWN: MPLS path PATH-BLACK-1-SEC down on LSP LSP-BLACK-1
May 1 19:48:01 2025 labdevice-RE1 rpd[33264]: RPD_MPLS_PATH_DOWN: MPLS path PATH-BLACK-2-SEC down on LSP LSP-BLACK-2
May 1 19:48:21 2025 labdevice-RE1 bfdd[30017]: BFDD_STATE_UP_TO_DOWN: BFD Session 172.26.13.1 (IFL 192) state Up -> Down LD/RD(2431/2545) Up time:10:05:07 Local diag: NbrSignal Remote diag: CtlExpire Reason: Received DOWN from PEER.
May 1 19:48:21 2025 labdevice-RE1 bfdd[30017]: BFDD_TRAP_SHOP_STATE_DOWN: local discriminator: 2431, new state: down, interface: et-3/0/6, peer addr: 172.26.13.1
May 1 19:48:23 2025 labdevice-RE1 bfdd[30017]: BFDD_TRAP_SHOP_STATE_UP: local discriminator: 2431, new state: up, interface: et-3/0/6, peer addr: 172.26.13.1
May 1 19:48:26 2025 labdevice-RE1 bfdd[30017]: BFDD_STATE_UP_TO_DOWN: BFD Session 172.26.13.1 (IFL 186) state Up -> Down LD/RD(2379/2478) Up time:01:53:09 Local diag: NbrSignal Remote diag: AdminDown Reason: Received ADMINDOWN from PEER.
May 1 19:48:26 2025 labdevice-RE1 bfdd[30017]: BFDD_TRAP_SHOP_STATE_DOWN: local discriminator: 2379, new state: down, interface: et-3/0/0, peer addr: 172.26.13.1
The flap is happening due to aggressive BFD timers. Legacy line cards LC1101 & LC1102 is not supporting inline micro-bfd sessions and “10 ms x 5” timer is aggressive for the ukern based bfd sessions.
BFD flaps can occur for these microbfd sessions when there is CPU load on the FPC. Before the event, we notice multiple MPLS path down. Any link down event or MPLS LSP path down can trigger CPU load on the fpc involved for processing state updates.
As the uBFD sessions are distributed to Microkernel (not inline to PFE ASIC), the minimum recommended timer is 100Ms X 3. In other words, uBFD is designed to operate as a PFE microkernel distributed session, not a PFE ASIC inline session that supports 10Ms timers.
If the uBFD sessions flaps even after configuring the recommended timers, below possibilities needs to be considered and investigated further.
Investigate the transmission side of the flapped AE member links, which are configured with 100ms X3.
If there were transmission events, like circuit switches, we can enable interface debug logs to prove that the transmission loss is crossing 300Ms whenever there is next hit happening.
If certain transmission paths record more than 300Ms loss, the customer needs to derive an optimum timer for those links like 100msX 4, 100msX5, or 100msXx, which makes the session stable.
Trace can be configured on both the BFD receiver and sender routers.
1.Enable PPM traceoptions
set routing-options ppm traceoptions file ppm size 25m count 10
set routing-options ppm traceoptions flag all
2.Enable BFD traceoptions
set protocols bfd traceoptions file bfd size 25m count 10
set protocols bfd traceoptions flag all
3.Enable interface traces in the pfe
>start shell pfe network fpc<slot>
debug interface flaps
debug interface micro-bfd
Enable millisecond pfe logging.
4. set system syslog file pfe pfe any
set system syslog time-format millisecond