L2Bridge link fails at STAR with “Interface already used for L2VPN”

Hi all,

I created a slice with a Tofino switch at STAR. One link failed because of what looks like a port conflict on star-data-sw. FABRIC then closed three more links, so the Tofino ended up with no working data-plane link.

Slice: ch3-tofino-star1026
Slice ID: c9c4fe15-37e1-4d9a-9494-ad9acba8b80d
Site: STAR
Lease: 25 Oct 19:30 UTC to 26 Oct 19:30 UTC
Final state: StableError

Topology: one Tofino switch (tofino1) and six VMs (h1, h4, h6, h7, s1, s5), joined by L2Bridge links.

What happened:

• All six VMs and the Tofino node reached Active.
• link-h1-s1 went to Failed. The NSO commit on star-data-sw was rejected with this error: “SEMANTIC ERRORS: l2vpn bridge group bg-link-h1-s1-… bridge-domain bd-link-h1-s1-… interface HundredGigE0/0/0/5.2020: Invalid argument: Interface already used for L2VPN”. The failed unit was 366de968-d015-40bb-ac0d-424b10fea718.
• link-tofino-s1, link-h4-tofino and link-h6-tofino were then Closed with: “TicketReviewPolicy: Closing reservation due to failure in slice”.
• Only link-h7-s5 and link-s5-s1 stayed Active, and neither goes through the Tofino.

Impact: the Tofino has no usable link to any VM. We can log into it, but we can’t send traffic through it. One working VM-to-Tofino link would be enough for our experiment, which only needs to generate traffic and read switch counters.

Questions:

1. Is HundredGigE0/0/0/5.2020 on star-data-sw still held by a stale L2VPN from another slice? Could that VLAN be freed?
2. Can a slice in StableError be modified? I would like to add a new link from a VM to an unused Tofino port without recreating the slice.
3. Is there a way to stop one failed link from closing the other links in the slice?

Thanks,
Haribabu K