Add: debug log for notion duplicate docs (#7079)

dust-tt · Sep 3, 2024 · f119b69 · f119b69
1 parent 3a5bf61
commit f119b69
Showing 1 changed file with 37 additions and 0 deletions.
diff --git a/connectors/src/connectors/notion/temporal/activities.ts b/connectors/src/connectors/notion/temporal/activities.ts
@@ -655,6 +655,17 @@ export async function garbageCollectorMarkAsSeen({
   const redisCli = await redisClient();
   const redisKey = redisGarbageCollectorKey(connector.id);
   if (pageIds.length > 0) {
+    // Temp: debug logging for a customer issue
+    const duplicatePageId = "167582ad-8d56-4f2d-a90c-bdb1289e55b6";
+    if (pageIds.find((id) => id === duplicatePageId)) {
+      localLogger.info(
+        {
+          pageIds: pageIds,
+        },
+        `Marking page id ${duplicatePageId} as seen`
+      );
+    }
+
     await redisCli.sAdd(`${redisKey}-pages`, pageIds);
   }
   if (databaseIds.length > 0) {
@@ -819,6 +830,21 @@ export async function garbageCollect({
       connector.id
     );
 
+    // Temp: debug logging for a customer issue
+    const duplicatePageId = "167582ad-8d56-4f2d-a90c-bdb1289e55b6";
+    if (
+      resourcesToCheck.find(
+        (r) => r.resourceType === "page" && r.resourceId === "duplicatePageId"
+      )
+    ) {
+      logger.info(
+        {
+          resourcesToCheck: resourcesToCheck,
+        },
+        `${duplicatePageId} is in the resources to check`
+      );
+    }
+
     const NOTION_UNHEALTHY_ERROR_CODES = [
       "internal_server_error",
       "notionhq_client_request_timeout",
@@ -1088,6 +1114,17 @@ async function findResourcesNotSeenInGarbageCollectionRun(
     return { pageIdsSeenInRun, databaseIdsSeenInRun };
   })();
 
+  // Temp: debug logging for a customer issue
+  const duplicatePageId = "167582ad-8d56-4f2d-a90c-bdb1289e55b6";
+  if (pageIdsSeenInRun.has(duplicatePageId)) {
+    logger.info(
+      {
+        pageIdsSeenInRun: pageIdsSeenInRun.entries(),
+      },
+      `${duplicatePageId} has been seen in this run`
+    );
+  }
+
   const pageSize = 500;
 
   const pagesNotSeenInGarbageCollectionRun: Array<{