Switching

Hello,

Do you have any idea how to resolve this issue?

I have two data-centers with same config, same qfx 5120- 48c model, same connections,junos OS. Bfd over bgp multi-hop
However, in one data-center, random bfd session are stuck in init or down state. Clearing bfd sessions doesn't resolve the issue.
When removing the BFD authentication , bfd sessions are going up.  Why in my second data-center the authentication is working and in the first one is not?
Evpn-vxlan spine-leaf



qfx 5120 48c
Junos os 20.1R3.9



----------traces
Feb 4 00:52:50 [THROTTLE]bfdd_rate_limit_program_timer_expiry: Session 172.31.100.3 is removed form program therad
Feb 4 00:52:50 bfdd_rate_limit_program_timer_expiry: Slow start revert
Feb 4 00:52:52 [RE_PPS]bfdd_slow_start_revert_poll_batch: RE PPS was 0 max_pps 5000 orginal_max_pps 5000 pend_pps 0 outstnading_poll 0


cli>show bfd session
Detect Transmit
Address State Interface Time Interval Multiplier
172.31.100.3 Down 6.000 2.000 3

show bfd session address 172.31.100.16
Detect Transmit
Address State Interface Time Interval Multiplier
172.31.100.16 Init 6.000 2.000 3


Regards,
Cristian

------------------------------
CRISTIAN CHIRATCU
------------------------------

This message was posted by a user wishing to remain anonymous

Resolution : Latest software upgrade

Original Message:
Sent: 02-03-2023 19:32
From: CRISTIAN CHIRATCU
Subject: random BFD sessions stuck in the Init or Down state

Hello,

Do you have any idea how to resolve this issue?

I have two data-centers with same config, same qfx 5120- 48c model, same connections,junos OS. Bfd over bgp multi-hop
However, in one data-center, random bfd session are stuck in init or down state. Clearing bfd sessions doesn't resolve the issue.
When removing the BFD authentication , bfd sessions are going up.  Why in my second data-center the authentication is working and in the first one is not?
Evpn-vxlan spine-leaf

qfx 5120 48c
Junos os 20.1R3.9



----------traces
Feb 4 00:52:50 [THROTTLE]bfdd_rate_limit_program_timer_expiry: Session 172.31.100.3 is removed form program therad
Feb 4 00:52:50 bfdd_rate_limit_program_timer_expiry: Slow start revert
Feb 4 00:52:52 [RE_PPS]bfdd_slow_start_revert_poll_batch: RE PPS was 0 max_pps 5000 orginal_max_pps 5000 pend_pps 0 outstnading_poll 0


cli>show bfd session
Detect Transmit
Address State Interface Time Interval Multiplier
172.31.100.3 Down 6.000 2.000 3

show bfd session address 172.31.100.16
Detect Transmit
Address State Interface Time Interval Multiplier
172.31.100.16 Init 6.000 2.000 3

Regards,
Cristian

------------------------------
CRISTIAN CHIRATCU
------------------------------

The QFX5120-48Y and -32C support Hardware-assisted Inline BFD that runs in the ASIC, but does not support authentication.  If auth is used, it will revert to normal Inline BFD, which runs in the PFE software. If the PFE is busy, it will not keep up with BFD polling and your sessions will fail. Try to increase the time to a few seconds and work your way down to see when the problems start.

"QFX5110, QFX5120, QFX5200, and QFX5210 switches support 10 multihop inline BFD sessions. You can configure them with a timer of 150 x 3 milliseconds. Single-hop sessions are also supported."

The above is valid if the PFE is not very busy of course. Do you really need 3 x 300 ms or is, for example, 4 x 600 ms acceptable?

https://www.juniper.net/documentation/us/en/software/junos/high-availability/topics/topic-map/bfd.html

/Fredrik

Original Message:
Sent: 02-03-2023 19:32
From: CRISTIAN CHIRATCU
Subject: random BFD sessions stuck in the Init or Down state

Hello,

Do you have any idea how to resolve this issue?

I have two data-centers with same config, same qfx 5120- 48c model, same connections,junos OS. Bfd over bgp multi-hop
However, in one data-center, random bfd session are stuck in init or down state. Clearing bfd sessions doesn't resolve the issue.
When removing the BFD authentication , bfd sessions are going up.  Why in my second data-center the authentication is working and in the first one is not?
Evpn-vxlan spine-leaf

qfx 5120 48c
Junos os 20.1R3.9



----------traces
Feb 4 00:52:50 [THROTTLE]bfdd_rate_limit_program_timer_expiry: Session 172.31.100.3 is removed form program therad
Feb 4 00:52:50 bfdd_rate_limit_program_timer_expiry: Slow start revert
Feb 4 00:52:52 [RE_PPS]bfdd_slow_start_revert_poll_batch: RE PPS was 0 max_pps 5000 orginal_max_pps 5000 pend_pps 0 outstnading_poll 0


cli>show bfd session
Detect Transmit
Address State Interface Time Interval Multiplier
172.31.100.3 Down 6.000 2.000 3

show bfd session address 172.31.100.16
Detect Transmit
Address State Interface Time Interval Multiplier
172.31.100.16 Init 6.000 2.000 3

Regards,
Cristian

------------------------------
CRISTIAN CHIRATCU
------------------------------

fixed by upgrading JUNOS OS.

------------------------------
CRISTIAN CHIRATCU
------------------------------

Original Message:
Sent: 07-02-2023 10:35
From: fb35523
Subject: random BFD sessions stuck in the Init or Down state

The QFX5120-48Y and -32C support Hardware-assisted Inline BFD that runs in the ASIC, but does not support authentication.  If auth is used, it will revert to normal Inline BFD, which runs in the PFE software. If the PFE is busy, it will not keep up with BFD polling and your sessions will fail. Try to increase the time to a few seconds and work your way down to see when the problems start.

"QFX5110, QFX5120, QFX5200, and QFX5210 switches support 10 multihop inline BFD sessions. You can configure them with a timer of 150 x 3 milliseconds. Single-hop sessions are also supported."

The above is valid if the PFE is not very busy of course. Do you really need 3 x 300 ms or is, for example, 4 x 600 ms acceptable?

https://www.juniper.net/documentation/us/en/software/junos/high-availability/topics/topic-map/bfd.html

/Fredrik

Original Message:
Sent: 02-03-2023 19:32
From: CRISTIAN CHIRATCU
Subject: random BFD sessions stuck in the Init or Down state

Hello,

Do you have any idea how to resolve this issue?

I have two data-centers with same config, same qfx 5120- 48c model, same connections,junos OS. Bfd over bgp multi-hop
However, in one data-center, random bfd session are stuck in init or down state. Clearing bfd sessions doesn't resolve the issue.
When removing the BFD authentication , bfd sessions are going up.  Why in my second data-center the authentication is working and in the first one is not?
Evpn-vxlan spine-leaf

qfx 5120 48c
Junos os 20.1R3.9



----------traces
Feb 4 00:52:50 [THROTTLE]bfdd_rate_limit_program_timer_expiry: Session 172.31.100.3 is removed form program therad
Feb 4 00:52:50 bfdd_rate_limit_program_timer_expiry: Slow start revert
Feb 4 00:52:52 [RE_PPS]bfdd_slow_start_revert_poll_batch: RE PPS was 0 max_pps 5000 orginal_max_pps 5000 pend_pps 0 outstnading_poll 0


cli>show bfd session
Detect Transmit
Address State Interface Time Interval Multiplier
172.31.100.3 Down 6.000 2.000 3

show bfd session address 172.31.100.16
Detect Transmit
Address State Interface Time Interval Multiplier
172.31.100.16 Init 6.000 2.000 3

Regards,
Cristian

------------------------------
CRISTIAN CHIRATCU
------------------------------