Gateway errors with .25, .26 and .27

[plush bluff]

Hey team, need some eyes on this.
Upgraded from stable 2026.4.23 to 2026.4.27 for native DeepSeek V4 Pro support.
openclaw gateway status showed:

Runtime: running (active)
Connectivity probe: ok
Discord provider logged in successfully
“ready” + heartbeat started

Yet no messages are delivered in Discord channels.
Logs show:

Repeated Gateway websocket closed: 1000
Gateway reconnect scheduled (zombie sessions)
slow-listener on INTERACTION_CREATE (durations 88s+)
stuck session + typing TTL reached (2m) → typing indicator dies
delivery-recovery failures with “Unknown Channel” even after setting groupPolicy: "open"

Tried: full doctor --fix, config overwrite with OPENCLAW_ALLOW_OLDER_BINARY_DESTRUCTIVE_ACTIONS=1, clean restart, etc. Same result.
Downgraded back to .23 — MiniMax flies again, zero disconnects, instant responses.
Anyone else hitting this on .25/.26/.27?
Is there a known regression in the Discord delivery pipeline or event queue after .24? Any workaround that doesn’t involve staying on .23?
Would really appreciate any insight before the next release drops. Thanks!

[vernal finch]

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

[night valley]

did you try local TUI? cause for me it's the same since upgrading but the gateway seems to be completely broken, either it's consuming a lot of CPU resources without working or it dies immediately