MPLS Kernel error 16.1.3.6

[Sergey_Vekli]

VPLS VPN not working. After disabling see error

MPLS gport resolve failed for unit = 0 vpn = 0x70e3, port = 0xffffffff vp = 0x18000a0c, rv = -7 (Entry not found)08/30/2016 11:37:25.99
extreme_mpls_del_vp_nh_xref: MPLS ECMP NH vp : 0xa0c l3intf = 0x83f map entry not found in btree rv = -7 (Entry not found)
extreme_custom_mpls_l2vpn_port_del: MPLS port delete: vp: 0xa0c l3Intf = 0x83f map entry not found in btree rv = -7 (Entry not found)
MPLS bcm_mpls_port_del failed for unit = 0 vpn = 0x70e3, vp = 0x18000a0c rv = -7 (Entry not found)


after enabling
* MPLS: bcm_mpls_port_add failed for unit = 0 vpn = 0x70e3, port = 0x8000809 vp = 0x18000a0d, rv = -14 (No resources for operation)

sh ver
Switch : Rev 12.0 BootROM: 2.0.2.1 IMG: 16.1.3.6

Image : ExtremeXOS version 16.1.3.6 16.1.3.6-patch1-9 by release-manager
on Mon Jul 18 10:45:49 EDT 2016
BootROM : 2.0.2.1
Diagnostics : 6.4

sh switch

System Type: X670-48x

Same problem resolved on previous release ?
https://extremeportal.force.com/ExtrArticleDetail?an=000063617

[Henrique]

Did you increase the BFD timer?

[Sergey_Vekli]

Ok. We enable log for bfd events.

[Henrique]

You can also add the BFD state change event in the log to track the flap as follows:

configure log filter DefaultFilter add events BFD.SessnStateChg

[Henrique]

I'm suspecting something with CPU and lower BFD timers since both OSPF neighbors configured with BFD have converged at same time as follows:

11/17/2016 12:22:57.57 Changing the state of neighbor rtid 10.15.0.1 ipa 10.16.248.18 to state = DOWN due to BFD path down.

11/17/2016 12:22:57.56 Changing the state of neighbor rtid 10.16.0.1 ipa 10.16.248.10 to state = DOWN due to BFD path down.

[Henrique]

Since the BFD is handled by the switch CPU, when using lower timers, if you have any CPU spike it could affect BFD.

If the issue occurs 3-4 times/day then you can try to adjust/increase the timer and check if the behavior changes.