From c3fe2a34665fac4389803a831905b1d9e8e2935f Mon Sep 17 00:00:00 2001
From: Logan Cusano <logan@simplestepsolutions.com>
Date: Tue, 18 Aug 2026 21:02:12 -0400
Subject: [PATCH] Page the org backfill instead of streaming whole collections

A bare .stream() over calls/ returned 503 "Query timed out. Please try either
limiting the entities scanned", which the Firestore client then re-raised as an
AttributeError from its own retry path -- so the real cause was only visible in
the chained traceback. The collection has simply outgrown a single scan.

Both passes now walk each collection in 500-document pages ordered by document
id, which needs no composite index. The counting pass also stops building a
list of every document just to count the ones missing org_id.

Documents written mid-run may be missed or seen twice; neither matters, since
post-tenancy code stamps org_id at write time and the update is idempotent, so
a second run cleans up anything the first skipped.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
---
 drb-c2-core/scripts/backfill_org_id.py | 48 ++++++++++++++++++++++----
 1 file changed, 42 insertions(+), 6 deletions(-)

diff --git a/drb-c2-core/scripts/backfill_org_id.py b/drb-c2-core/scripts/backfill_org_id.py
index e130d97..f15fcda 100644
--- a/drb-c2-core/scripts/backfill_org_id.py
+++ b/drb-c2-core/scripts/backfill_org_id.py
@@ -76,6 +76,38 @@ TENANT_COLLECTIONS = ["nodes", "systems", "calls", "incidents", "alert_rules", "
 _BATCH_SIZE = 400
 
 
+_PAGE_SIZE = 500
+
+
+def _iter_collection(db, collection: str, page_size: int = _PAGE_SIZE):
+    """
+    Walk a collection in key-ordered pages instead of one open stream.
+
+    A bare .stream() over calls/ died with "503 Query timed out. Please try
+    either limiting the entities scanned" -- the collection has outgrown what
+    Firestore will serve as a single scan, and the failure surfaced as an
+    unrelated-looking AttributeError from the client's own retry path. Paging by
+    document id needs no composite index and keeps each request small; it also
+    stops the counting pass holding every document in memory at once.
+
+    Documents written while this runs may be missed or seen twice. Neither
+    matters: post-tenancy code stamps org_id itself, and the update below is
+    idempotent.
+    """
+    base = db.collection(collection).order_by("__name__").limit(page_size)
+    cursor = None
+    while True:
+        query = base.start_after(cursor) if cursor is not None else base
+        page = list(query.stream())
+        if not page:
+            return
+        for doc in page:
+            yield doc
+        if len(page) < page_size:
+            return
+        cursor = page[-1]
+
+
 def main() -> None:
     parser = argparse.ArgumentParser(
         description=__doc__,
@@ -126,11 +158,15 @@ def main() -> None:
     counts: dict[str, tuple[int, int]] = {}
     total_missing = 0
     for collection in TENANT_COLLECTIONS:
-        docs = list(db.collection(collection).stream())
-        missing = [d for d in docs if not (d.to_dict() or {}).get("org_id")]
-        counts[collection] = (len(docs), len(missing))
-        total_missing += len(missing)
-        print(f"{collection}: {len(docs)} docs total, {len(missing)} missing org_id")
+        total = 0
+        missing = 0
+        for doc in _iter_collection(db, collection):
+            total += 1
+            if not (doc.to_dict() or {}).get("org_id"):
+                missing += 1
+        counts[collection] = (total, missing)
+        total_missing += missing
+        print(f"{collection}: {total} docs total, {missing} missing org_id")
 
     print(f"\nTotal documents to backfill: {total_missing}")
 
@@ -179,7 +215,7 @@ def main() -> None:
         batch = db.batch()
         batch_count = 0
         written = 0
-        for doc in db.collection(collection).stream():
+        for doc in _iter_collection(db, collection):
             if (doc.to_dict() or {}).get("org_id"):
                 continue
             batch.update(doc.reference, {"org_id": FOUNDING_ORG_ID})