Lightning gateway sends consistently failing on signet (~3min timeout then revoke)

Over the course of testing tonight (2026-06-21, roughly 19:00–20:45 UTC), I hit a consistent failure pattern on Lightning-gateway sends against ark.signet.2nd.dev, while Ark-native (arkoor) operations worked perfectly throughout.

Pattern observed:

4 consecutive Lightning send attempts failed, each taking ~100–200 seconds before resolving to a failed/canceled state and auto-revoking

Failures occurred in both directions between two different wallets (bark 0.2.5 CLI)

Failures occurred on both CLI-generated invoices (bark lightning invoice) and invoices generated via the bark-wallet Rust library directly — ruling out anything client-specific

Amounts tested: 10–1000 sats

Every arkoor (Ark-to-Ark) send/receive during the same session succeeded instantly (sub-second) with 0 fee, no issues

Representative failure log:

[INFO] Sending bolt11 payment of 0.00001000 BTC to invoice lntbs10u1p4rs…

[INFO] Lightning payment failed, preparing to revoke

[INFO] Payment completed: hash =

~108–195 seconds elapsed between send and failure across the 4 attempts.

Notable secondary observation: in one case, a receiving wallet logged “Lightning receive claimed via mailbox notification” for a payment hash, immediately followed by “payment was canceled. removing pending lightning receive” within ~40ms — suggesting a possible race between claim and cancellation rather than a clean upfront routing failure.

Also observed (separately, may be related or may be a distinct issue): a fresh wallet’s first Lightning receive failed with vtxopool error: More than one change output when the wallet had zero prior VTXOs. Pre-seeding the wallet with one Ark-native deposit before attempting a Lightning receive resolved this on subsequent attempts. Happy to file this separately if useful.

Environment: bark 0.2.5, both CLI and bark-wallet crate direct, against ark.signet.2nd.dev / esplora.signet.2nd.dev.

Wanted to flag in case this is a known liquidity-tuning issue on the signet gateway, or something worth a closer look. Happy to provide full logs/datadir details if useful.

1 Like

Thanks for the report @Codepocketdev . We’re looking into it.

Hey! We looked into this and it turned out we had some problem in our faucet that should be resolved now. Lightning receives from the faucet should be working again.

Thanks for reporting!