Fix migration code in spite of conflicts

Author: ChrisPenner

sql/2025-01-24_migrate-serialized-entities.sql

@@ -67,6 +67,40 @@ CREATE TABLE migrate_serialized_queue_unsandboxed (
 
 -- EXTRA
 
+-- If some causals are missing:
+-- INSERT INTO migrate_serialized_queue_unsandboxed (hash, user_id)
+--     SELECT DISTINCT ON(c.hash) c.hash, co.user_id
+--       FROM causal_ownership co
+--         JOIN causals c ON co.causal_id = c.id
+--   WHERE NOT EXISTS(SELECT FROM serialized_causals sc WHERE sc.causal_id = c.id)
+--   ;
+
+
+-- If some namespaces are missing:
+-- INSERT INTO migrate_serialized_queue_unsandboxed (hash, user_id)
+--     SELECT DISTINCT ON(bh.base32) bh.base32, no.user_id
+--       FROM namespace_ownership no
+--         JOIN branch_hashes bh ON no.namespace_hash_id = bh.id
+--   WHERE NOT EXISTS(SELECT FROM serialized_namespaces sn WHERE sn.namespace_hash_id = bh.id)
+--   ;
+
+-- If some term components are missing
+-- INSERT INTO migrate_serialized_queue_sandboxed (component_hash_id, user_id)
+-- SELECT DISTINCT ON (t.component_hash_id) t.component_hash_id, st.user_id
+--   FROM sandboxed_terms st
+--     JOIN terms t ON st.term_id = t.id
+--   WHERE NOT EXISTS(SELECT FROM serialized_components sc WHERE sc.user_id = st.user_id AND sc.component_hash_id = t.component_hash_id)
+--   ;
+
+-- If some type components are missing
+-- INSERT INTO migrate_serialized_queue_sandboxed (component_hash_id, user_id)
+-- SELECT DISTINCT ON (t.component_hash_id) t.component_hash_id, st.user_id
+--   FROM sandboxed_types st
+--     JOIN types t ON st.type_id = t.id
+--   WHERE NOT EXISTS(SELECT FROM serialized_components sc WHERE sc.user_id = st.user_id AND sc.component_hash_id = t.component_hash_id)
+--   ;
+
+
 -- If some components were somehow missing, this will detect which ones still need processing and add them back to
 -- the queue.
 -- INSERT INTO migrate_serialized_queue_sandboxed (component_hash_id, user_id)

src/Share/BackgroundJobs/SerializedEntitiesMigration/Worker.hs

@@ -65,20 +65,21 @@ processComponents !_authZReceipt = do
     Q.claimComponent >>= \case
       Nothing -> pure False
       Just (componentHashId, userId) -> do
-            let codebaseEnv = CodebaseEnv userId
-            Codebase.codebaseMToTransaction codebaseEnv $ do
-              hash32 <- (Hash32.fromHash . unComponentHash) <$> HQ.expectComponentHashesOf id componentHashId
-              entity <- SQ.expectEntity hash32
-              componentSummaryDigest <- HQ.expectComponentSummaryDigest componentHashId
-              let tempEntity = SyncCommon.entityToTempEntity id entity
-              let (SyncV2.CBORBytes bytes) = SyncV2.serialiseCBORBytes tempEntity
-              bytesId <- DefnQ.ensureBytesIdsOf id (BL.toStrict bytes)
-              execute_
-                [sql|
+        let codebaseEnv = CodebaseEnv userId
+        Codebase.codebaseMToTransaction codebaseEnv $ do
+          hash32 <- (Hash32.fromHash . unComponentHash) <$> HQ.expectComponentHashesOf id componentHashId
+          entity <- SQ.expectEntity hash32
+          componentSummaryDigest <- HQ.expectComponentSummaryDigest componentHashId
+          let tempEntity = SyncCommon.entityToTempEntity id entity
+          let (SyncV2.CBORBytes bytes) = SyncV2.serialiseCBORBytes tempEntity
+          bytesId <- DefnQ.ensureBytesIdsOf id (BL.toStrict bytes)
+          execute_
+            [sql|
                 INSERT INTO component_summary_digests_to_serialized_component_bytes_hash (component_hash_id, component_summary_digest, serialized_component_bytes_id)
                   VALUES (#{componentHashId}, #{componentSummaryDigest}, #{bytesId})
+                  ON CONFLICT DO NOTHING
                 |]
-              pure True
+          pure True
 
 saveUnsandboxedSerializedEntities :: (QueryM m) => Hash32 -> TempEntity -> m ()
 saveUnsandboxedSerializedEntities hash entity = do