diff --git a/swh/storage/migrate_extrinsic_metadata.py b/swh/storage/migrate_extrinsic_metadata.py
--- a/swh/storage/migrate_extrinsic_metadata.py
+++ b/swh/storage/migrate_extrinsic_metadata.py
@@ -27,10 +27,12 @@
 import os
 import re
 import sys
+import time
 from typing import Any, Dict, Optional
 from urllib.parse import unquote, urlparse
 
 import iso8601
+import psycopg2
 
 from swh.core.db import BaseDb
 from swh.model.hashutil import hash_to_hex
@@ -41,6 +43,7 @@
     MetadataFetcher,
     MetadataTargetType,
     RawExtrinsicMetadata,
+    Sha1Git,
 )
 from swh.storage import get_storage
 
@@ -832,6 +835,38 @@
         )
 
 
+def iter_revision_rows(storage_dbconn: str, first_id: Sha1Git):
+    after_id = first_id
+    failures = 0
+    while True:
+        try:
+            storage_db = BaseDb.connect(storage_dbconn)
+            with storage_db.cursor() as cur:
+                while True:
+                    cur.execute(
+                        f"SELECT {', '.join(REVISION_COLS)} FROM revision "
+                        f"WHERE id > %s AND metadata IS NOT NULL "
+                        f"ORDER BY id LIMIT 1000",
+                        (after_id,),
+                    )
+                    new_rows = 0
+                    for row in cur:
+                        new_rows += 1
+                        row_d = dict(zip(REVISION_COLS, row))
+                        yield row_d
+                    after_id = row_d["id"]
+                    if new_rows == 0:
+                        return
+        except psycopg2.OperationalError as e:
+            print(e)
+            # most likely a temporary error, try again
+            if failures >= 60:
+                raise
+            else:
+                time.sleep(60)
+                failures += 1
+
+
 def main(storage_dbconn, storage_url, deposit_dbconn, first_id, dry_run):
     storage_db = BaseDb.connect(storage_dbconn)
     deposit_db = BaseDb.connect(deposit_dbconn)
@@ -845,33 +880,19 @@
         # authorities that differ from what the loaders use.
 
     total_rows = 0
-    with storage_db.cursor() as read_cur:
-        with deposit_db.cursor() as deposit_cur:
-            after_id = first_id
-            while True:
-                read_cur.execute(
-                    f"SELECT {', '.join(REVISION_COLS)} FROM revision "
-                    f"WHERE id > %s AND metadata IS NOT NULL ORDER BY id LIMIT 1000",
-                    (after_id,),
-                )
-                new_rows = 0
-                for row in read_cur:
-                    row_d = dict(zip(REVISION_COLS, row))
-                    handle_row(row_d, storage, deposit_cur, dry_run)
-                    new_rows += 1
-
-                if new_rows == 0:
-                    break
+    with deposit_db.cursor() as deposit_cur:
+        for row in iter_revision_rows(storage_dbconn, first_id):
+            handle_row(row, storage, deposit_cur, dry_run)
 
-                after_id = row_d["id"]
+            total_rows += 1
 
-                total_rows += new_rows
+            if total_rows % 1000 == 0:
                 percents = (
-                    int.from_bytes(after_id[0:4], byteorder="big") * 100 / (1 << 32)
+                    int.from_bytes(row["id"][0:4], byteorder="big") * 100 / (1 << 32)
                 )
                 print(
-                    f"Migrated {total_rows/1000000.:.2f}M rows "
-                    f"(~{percents:.1f}%, last revision: {after_id.hex()})"
+                    f"Processed {total_rows/1000000.:.2f}M rows "
+                    f"(~{percents:.1f}%, last revision: {row['id'].hex()})"
                 )