From 5dd19edb5661f74c3edc905c8c677ae54e82a26c Mon Sep 17 00:00:00 2001
From: Owen <owen@txv.io>
Date: Mon, 1 Jun 2026 14:05:19 -0700
Subject: [PATCH] Hold the hp error message until after 18 tries

---
 server/routers/olm/error.ts                   |  2 +-
 .../routers/olm/handleOlmRegisterMessage.ts   | 55 ++++++++++++++++++-
 2 files changed, 55 insertions(+), 2 deletions(-)

diff --git a/server/routers/olm/error.ts b/server/routers/olm/error.ts
index 15047b94c..d9058a3f6 100644
--- a/server/routers/olm/error.ts
+++ b/server/routers/olm/error.ts
@@ -93,7 +93,7 @@ export const OlmErrorCodes = {
     },
     HOLEPUNCH_MISSING: {
         code: "HOLEPUNCH_MISSING",
-        message: `Hole punch information is missing. Please ensure your client can reach the server on UDP port ${udpPort} and try registering again.`
+        message: `Unable to coordinate client P2P connection. Please ensure your client can reach the server on UDP port ${udpPort} and try registering again.`
     }
 } as const;
 
diff --git a/server/routers/olm/handleOlmRegisterMessage.ts b/server/routers/olm/handleOlmRegisterMessage.ts
index 747770bff..3b0e1637a 100644
--- a/server/routers/olm/handleOlmRegisterMessage.ts
+++ b/server/routers/olm/handleOlmRegisterMessage.ts
@@ -20,6 +20,14 @@ import { handleFingerprintInsertion } from "./fingerprintingUtils";
 import { build } from "@server/build";
 import { canCompress } from "@server/lib/clientVersionChecks";
 import config from "@server/lib/config";
+import cache from "#dynamic/lib/cache";
+
+const HOLEPUNCH_STALE_CHAIN_THRESHOLD = 18;
+const HOLEPUNCH_STALE_CHAIN_TTL_SECONDS = 1800;
+
+function getHolePunchChainCounterKey(olmId: string, chainId: string): string {
+    return `olm:register:stale_holepunch:${olmId}:${chainId}`;
+}
 
 export const handleOlmRegisterMessage: MessageHandler = async (context) => {
     logger.info("[handleOlmRegisterMessage] Handling register olm message");
@@ -319,6 +327,24 @@ export const handleOlmRegisterMessage: MessageHandler = async (context) => {
             );
     }
 
+    let staleHolePunchChainCount: number | undefined;
+    const hasChainId =
+        chainId !== undefined && chainId !== null && String(chainId) !== "";
+
+    if (hasChainId) {
+        const cacheKey = getHolePunchChainCounterKey(
+            olm.olmId,
+            String(chainId)
+        );
+        const existingCount = (await cache.get<number>(cacheKey)) ?? 0;
+        staleHolePunchChainCount = existingCount + 1;
+        await cache.set(
+            cacheKey,
+            staleHolePunchChainCount,
+            HOLEPUNCH_STALE_CHAIN_TTL_SECONDS
+        );
+    }
+
     // this prevents us from accepting a register from an olm that has not hole punched yet.
     // the olm will pump the register so we can keep checking
     // TODO: I still think there is a better way to do this rather than locking it out here but ???
@@ -327,7 +353,34 @@ export const handleOlmRegisterMessage: MessageHandler = async (context) => {
             `[handleOlmRegisterMessage] Client last hole punch is too old and we have sites to send; skipping this register. The client is failing to hole punch and identify its network address with the server. Can the client reach the server on UDP port ${config.getRawConfig().gerbil.clients_start_port}?`,
             { orgId: client.orgId, clientId: client.clientId }
         );
-        sendOlmError(OlmErrorCodes.HOLEPUNCH_MISSING, olm.olmId);
+
+        if (!hasChainId) {
+            logger.debug(
+                "[handleOlmRegisterMessage] Skipping HOLEPUNCH_MISSING because chainId is missing",
+                {
+                    orgId: client.orgId,
+                    clientId: client.clientId,
+                    olmId: olm.olmId
+                }
+            );
+            return;
+        }
+
+        if (staleHolePunchChainCount === HOLEPUNCH_STALE_CHAIN_THRESHOLD) {
+            sendOlmError(OlmErrorCodes.HOLEPUNCH_MISSING, olm.olmId);
+        } else {
+            logger.debug(
+                "[handleOlmRegisterMessage] Suppressing HOLEPUNCH_MISSING until chain threshold is met",
+                {
+                    orgId: client.orgId,
+                    clientId: client.clientId,
+                    olmId: olm.olmId,
+                    chainId,
+                    staleHolePunchChainCount,
+                    threshold: HOLEPUNCH_STALE_CHAIN_THRESHOLD
+                }
+            );
+        }
         return;
     }