[indexer] merge in indexer breaking change park (#18899)

## Description 

Changes in this uhaul PR includes:
- Addition of `objects_version` table
- Addition of a number of transaction and event indices tables, with the
removal of `tx_calls` table.
- Enable pruning for the newly added indices tables and make tx sequence
number a valid partition range, in addition to cp sequence number.

## Test plan 

tested locally

---

## Release notes

This PR modifies the indexer db schemas for improving GraphQL query
performance. Specifically, an `objects_version` table along with various
transaction and events lookup tables are added. `tx_calls` table is
removed by more fine-grained tables `tx_calls_pkg`, `tx_calls_mod` and
`tx_calls_fun`.

- [ ] Protocol: 
- [ ] Nodes (Validators and Full nodes): 
- [x] Indexer: 
- [ ] JSON-RPC: 
- [ ] GraphQL: 
- [ ] CLI: 
- [ ] Rust SDK:
- [ ] REST API:

---------

Co-authored-by: Ashok Menon <ashok@mystenlabs.com>
Co-authored-by: wlmyng <127570466+wlmyng@users.noreply.github.com>

crates/sui-graphql-rpc/src/types/transaction_block.rs

@@ -8,14 +8,14 @@ use async_graphql::{
     dataloader::Loader,
     *,
 };
-use diesel::{BoolExpressionMethods, ExpressionMethods, JoinOnDsl, QueryDsl, SelectableHelper};
+use diesel::{ExpressionMethods, JoinOnDsl, QueryDsl, SelectableHelper};
 use fastcrypto::encoding::{Base58, Encoding};
 use serde::{Deserialize, Serialize};
 use sui_indexer::{
     models::transactions::StoredTransaction,
     schema::{
-        transactions, tx_calls, tx_changed_objects, tx_digests, tx_input_objects, tx_recipients,
-        tx_senders,
+        transactions, tx_calls_fun, tx_changed_objects, tx_digests, tx_input_objects,
+        tx_recipients, tx_senders,
     },
 };
 use sui_types::{
@@ -318,15 +318,15 @@ impl TransactionBlock {
                         let mut query = tx::dsl::transactions.into_boxed();
 
                         if let Some(f) = &filter.function {
-                            let sub_query = tx_calls::dsl::tx_calls
-                                .select(tx_calls::dsl::tx_sequence_number)
+                            let sub_query = tx_calls_fun::dsl::tx_calls_fun
+                                .select(tx_calls_fun::dsl::tx_sequence_number)
                                 .into_boxed();
 
                             query = query.filter(tx::dsl::tx_sequence_number.eq_any(f.apply(
                                 sub_query,
-                                tx_calls::dsl::package,
-                                tx_calls::dsl::module,
-                                tx_calls::dsl::func,
+                                tx_calls_fun::dsl::package,
+                                tx_calls_fun::dsl::module,
+                                tx_calls_fun::dsl::func,
                             )));
                         }
 
@@ -507,9 +507,7 @@ impl Loader<DigestKey> for Db {
         let transactions: Vec<StoredTransaction> = self
             .execute(move |conn| {
                 conn.results(move || {
-                    let join = ds::cp_sequence_number
-                        .eq(tx::checkpoint_sequence_number)
-                        .and(ds::tx_sequence_number.eq(tx::tx_sequence_number));
+                    let join = ds::tx_sequence_number.eq(tx::tx_sequence_number);
 
                     tx::transactions
                         .inner_join(ds::tx_digests.on(join))

crates/sui-indexer/migrations/mysql/2024-04-24-180249_packages/up.sql

@@ -1,7 +1,14 @@
 CREATE TABLE packages
 (
-    package_id                   blob          NOT NULL,
+    package_id                   BLOB          NOT NULL,
+    original_id                  BLOB          NOT NULL,
+    package_version              BIGINT        NOT NULL,
     -- bcs serialized MovePackage
-    move_package                 MEDIUMBLOB          NOT NULL,
-        CONSTRAINT packages_pk PRIMARY KEY (package_id(255))
+    move_package                 MEDIUMBLOB    NOT NULL,
+    checkpoint_sequence_number   BIGINT        NOT NULL,
+    CONSTRAINT packages_pk PRIMARY KEY (package_id(32), original_id(32), package_version),
+    CONSTRAINT packages_unique_package_id UNIQUE (package_id(32))
 );
+
+CREATE INDEX packages_cp_id_version ON packages (checkpoint_sequence_number, original_id(32), package_version);
+CREATE INDEX packages_id_version_cp ON packages (original_id(32), package_version, checkpoint_sequence_number);

crates/sui-indexer/migrations/mysql/2024-05-05-155158_obj_indices/down.sql

@@ -0,0 +1 @@
+DROP TABLE IF EXISTS objects_version;

crates/sui-indexer/migrations/mysql/2024-05-05-155158_obj_indices/up.sql

@@ -0,0 +1,9 @@
+-- The Postgres version of this table is partitioned by the first byte
+-- of object_id, but this kind of partition is not easily supported in
+-- MySQL, so this variant is unpartitioned for now.
+CREATE TABLE objects_version (
+    object_id           BLOB          NOT NULL,
+    object_version      BIGINT        NOT NULL,
+    cp_sequence_number  BIGINT        NOT NULL,
+    PRIMARY KEY (object_id(32), object_version)
+)

crates/sui-indexer/migrations/pg/2023-08-19-044020_events/up.sql

@@ -1,3 +1,4 @@
+-- TODO: modify queries in indexer reader to take advantage of the new indices
 CREATE TABLE events
 (
     tx_sequence_number          BIGINT       NOT NULL,
@@ -23,8 +24,8 @@ CREATE TABLE events
     timestamp_ms                BIGINT       NOT NULL,
     -- bcs of the Event contents (Event.contents)
     bcs                         BYTEA        NOT NULL,
-    PRIMARY KEY(tx_sequence_number, event_sequence_number, checkpoint_sequence_number)
-) PARTITION BY RANGE (checkpoint_sequence_number);
+    PRIMARY KEY(tx_sequence_number, event_sequence_number)
+) PARTITION BY RANGE (tx_sequence_number);
 CREATE TABLE events_partition_0 PARTITION OF events FOR VALUES FROM (0) TO (MAXVALUE);
 CREATE INDEX events_package ON events (package, tx_sequence_number, event_sequence_number);
 CREATE INDEX events_package_module ON events (package, module, tx_sequence_number, event_sequence_number);

crates/sui-indexer/migrations/pg/2023-08-19-044026_transactions/down.sql

@@ -1,2 +1,3 @@
 -- This file should undo anything in `up.sql`
 DROP TABLE IF EXISTS transactions;
+DROP TABLE IF EXISTS transactions_partition_0;

crates/sui-indexer/migrations/pg/2023-08-19-044026_transactions/up.sql

@@ -18,10 +18,6 @@ CREATE TABLE transactions (
     -- number of successful commands in this transaction, bound by number of command
     -- in a programmaable transaction.
     success_command_count       smallint     NOT NULL,
-    PRIMARY KEY (tx_sequence_number, checkpoint_sequence_number)
-) PARTITION BY RANGE (checkpoint_sequence_number);
+    PRIMARY KEY (tx_sequence_number)
+) PARTITION BY RANGE (tx_sequence_number);
 CREATE TABLE transactions_partition_0 PARTITION OF transactions FOR VALUES FROM (0) TO (MAXVALUE);
-CREATE INDEX transactions_transaction_digest ON transactions (transaction_digest);
-CREATE INDEX transactions_checkpoint_sequence_number ON transactions (checkpoint_sequence_number);
--- only create index for system transactions (0). See types.rs
-CREATE INDEX transactions_transaction_kind ON transactions (transaction_kind) WHERE transaction_kind = 0;

crates/sui-indexer/migrations/pg/2023-08-19-044044_checkpoints/up.sql

@@ -1,27 +1,29 @@
 CREATE TABLE checkpoints
 (
-    sequence_number                     bigint       PRIMARY KEY,
-    checkpoint_digest                   bytea        NOT NULL,
-    epoch                               bigint       NOT NULL,
+    sequence_number                     BIGINT       PRIMARY KEY,
+    checkpoint_digest                   BYTEA        NOT NULL,
+    epoch                               BIGINT       NOT NULL,
     -- total transactions in the network at the end of this checkpoint (including itself)
-    network_total_transactions          bigint       NOT NULL,
-    previous_checkpoint_digest          bytea,
+    network_total_transactions          BIGINT       NOT NULL,
+    previous_checkpoint_digest          BYTEA,
     -- if this checkpoitn is the last checkpoint of an epoch
     end_of_epoch                        boolean      NOT NULL,
     -- array of TranscationDigest in bytes included in this checkpoint
-    tx_digests                          bytea[]      NOT NULL,
+    tx_digests                          BYTEA[]      NOT NULL,
     timestamp_ms                        BIGINT       NOT NULL,
     total_gas_cost                      BIGINT       NOT NULL,
     computation_cost                    BIGINT       NOT NULL,
     storage_cost                        BIGINT       NOT NULL,
     storage_rebate                      BIGINT       NOT NULL,
     non_refundable_storage_fee          BIGINT       NOT NULL,
     -- bcs serialized Vec<CheckpointCommitment> bytes
-    checkpoint_commitments              bytea        NOT NULL,
+    checkpoint_commitments              BYTEA        NOT NULL,
     -- bcs serialized AggregateAuthoritySignature bytes
-    validator_signature                 bytea        NOT NULL,
+    validator_signature                 BYTEA        NOT NULL,
     -- bcs serialzied EndOfEpochData bytes, if the checkpoint marks end of an epoch
-    end_of_epoch_data                   bytea
+    end_of_epoch_data                   BYTEA,
+    min_tx_sequence_number              BIGINT,
+    max_tx_sequence_number              BIGINT
 );
 
 CREATE INDEX checkpoints_epoch ON checkpoints (epoch, sequence_number);

crates/sui-indexer/migrations/pg/2023-08-19-060729_packages/up.sql

@@ -1,6 +1,14 @@
-CREATE TABLE packages 
+CREATE TABLE packages
 (
-    package_id                   bytea          PRIMARY KEY,
+    package_id                   bytea          NOT NULL,
+    original_id                  bytea          NOT NULL,
+    package_version              bigint         NOT NULL,
     -- bcs serialized MovePackage
-    move_package                 bytea          NOT NULL
+    move_package                 bytea          NOT NULL,
+    checkpoint_sequence_number   bigint         NOT NULL,
+    CONSTRAINT packages_pkey PRIMARY KEY (package_id, original_id, package_version),
+    CONSTRAINT packages_unique_package_id UNIQUE (package_id)
 );
+
+CREATE INDEX packages_cp_id_version ON packages (checkpoint_sequence_number, original_id, package_version);
+CREATE INDEX packages_id_version_cp ON packages (original_id, package_version, checkpoint_sequence_number);

crates/sui-indexer/migrations/pg/2023-10-06-204335_tx_indices/down.sql

@@ -2,5 +2,8 @@ DROP TABLE IF EXISTS tx_senders;
 DROP TABLE IF EXISTS tx_recipients;
 DROP TABLE IF EXISTS tx_input_objects;
 DROP TABLE IF EXISTS tx_changed_objects;
-DROP TABLE IF EXISTS tx_calls;
+DROP TABLE IF EXISTS tx_calls_pkg;
+DROP TABLE IF EXISTS tx_calls_mod;
+DROP TABLE IF EXISTS tx_calls_fun;
 DROP TABLE IF EXISTS tx_digests;
+DROP TABLE IF EXISTS tx_kinds;

crates/sui-indexer/migrations/pg/2023-10-06-204335_tx_indices/up.sql

@@ -1,57 +1,70 @@
 CREATE TABLE tx_senders (
-    cp_sequence_number          BIGINT       NOT NULL,
     tx_sequence_number          BIGINT       NOT NULL,
-    -- SuiAddress in bytes.
     sender                      BYTEA        NOT NULL,
-    PRIMARY KEY(sender, tx_sequence_number, cp_sequence_number)
+    PRIMARY KEY(sender, tx_sequence_number)
 );
-CREATE INDEX tx_senders_tx_sequence_number_index ON tx_senders (tx_sequence_number, cp_sequence_number);
 
 CREATE TABLE tx_recipients (
-    cp_sequence_number          BIGINT       NOT NULL,
     tx_sequence_number          BIGINT       NOT NULL,
-    -- SuiAddress in bytes.
     recipient                   BYTEA        NOT NULL,
-    PRIMARY KEY(recipient, tx_sequence_number, cp_sequence_number)
+    sender                      BYTEA        NOT NULL,
+    PRIMARY KEY(recipient, tx_sequence_number)
 );
-CREATE INDEX tx_recipients_tx_sequence_number_index ON tx_recipients (tx_sequence_number, cp_sequence_number);
+CREATE INDEX tx_recipients_sender ON tx_recipients (sender, recipient, tx_sequence_number);
 
 CREATE TABLE tx_input_objects (
-    cp_sequence_number          BIGINT       NOT NULL,
     tx_sequence_number          BIGINT       NOT NULL,
-    -- Object ID in bytes. 
     object_id                   BYTEA        NOT NULL,
-    PRIMARY KEY(object_id, tx_sequence_number, cp_sequence_number)
+    sender                      BYTEA        NOT NULL,
+    PRIMARY KEY(object_id, tx_sequence_number)
 );
 CREATE INDEX tx_input_objects_tx_sequence_number_index ON tx_input_objects (tx_sequence_number);
+CREATE INDEX tx_input_objects_sender ON tx_input_objects (sender, object_id, tx_sequence_number);
 
 CREATE TABLE tx_changed_objects (
-    cp_sequence_number          BIGINT       NOT NULL,
     tx_sequence_number          BIGINT       NOT NULL,
-    -- Object Id in bytes.
     object_id                   BYTEA        NOT NULL,
-    PRIMARY KEY(object_id, tx_sequence_number, cp_sequence_number)
+    sender                      BYTEA        NOT NULL,
+    PRIMARY KEY(object_id, tx_sequence_number)
 );
 CREATE INDEX tx_changed_objects_tx_sequence_number_index ON tx_changed_objects (tx_sequence_number);
+CREATE INDEX tx_changed_objects_sender ON tx_changed_objects (sender, object_id, tx_sequence_number);
+
+CREATE TABLE tx_calls_pkg (
+    tx_sequence_number          BIGINT       NOT NULL,
+    package                     BYTEA        NOT NULL,
+    sender                      BYTEA        NOT NULL,
+    PRIMARY KEY(package, tx_sequence_number)
+);
+CREATE INDEX tx_calls_pkg_sender ON tx_calls_pkg (sender, package, tx_sequence_number);
+
+CREATE TABLE tx_calls_mod (
+    tx_sequence_number          BIGINT       NOT NULL,
+    package                     BYTEA        NOT NULL,
+    module                      TEXT         NOT NULL,
+    sender                      BYTEA        NOT NULL,
+    PRIMARY KEY(package, module, tx_sequence_number)
+);
+CREATE INDEX tx_calls_mod_sender ON tx_calls_mod (sender, package, module, tx_sequence_number);
 
-CREATE TABLE tx_calls (
-    cp_sequence_number          BIGINT       NOT NULL,
+CREATE TABLE tx_calls_fun (
     tx_sequence_number          BIGINT       NOT NULL,
     package                     BYTEA        NOT NULL,
     module                      TEXT         NOT NULL,
     func                        TEXT         NOT NULL,
-    -- 1. Using Primary Key as a unique index.
-    -- 2. Diesel does not like tables with no primary key.
-    PRIMARY KEY(package, tx_sequence_number, cp_sequence_number)
+    sender                      BYTEA        NOT NULL,
+    PRIMARY KEY(package, module, func, tx_sequence_number)
 );
-CREATE INDEX tx_calls_module ON tx_calls (package, module, tx_sequence_number, cp_sequence_number);
-CREATE INDEX tx_calls_func ON tx_calls (package, module, func, tx_sequence_number, cp_sequence_number);
-CREATE INDEX tx_calls_tx_sequence_number ON tx_calls (tx_sequence_number, cp_sequence_number);
+CREATE INDEX tx_calls_fun_sender ON tx_calls_fun (sender, package, module, func, tx_sequence_number);
 
--- un-partitioned table for tx_digest -> (cp_sequence_number, tx_sequence_number) lookup.
 CREATE TABLE tx_digests (
     tx_digest                   BYTEA        PRIMARY KEY,
-    cp_sequence_number          BIGINT       NOT NULL,
     tx_sequence_number          BIGINT       NOT NULL
 );
 CREATE INDEX tx_digests_tx_sequence_number ON tx_digests (tx_sequence_number);
+
+CREATE TABLE tx_kinds (
+    tx_sequence_number          BIGINT       NOT NULL,
+    tx_kind                     SMALLINT     NOT NULL,
+    PRIMARY KEY(tx_kind, tx_sequence_number)
+);

crates/sui-indexer/migrations/pg/2023-11-29-193859_advance_partition/up.sql

@@ -1,10 +1,10 @@
-CREATE OR REPLACE PROCEDURE advance_partition(table_name TEXT, last_epoch BIGINT, new_epoch BIGINT, last_epoch_start_cp BIGINT, new_epoch_start_cp BIGINT)
+CREATE OR REPLACE PROCEDURE advance_partition(table_name TEXT, last_epoch BIGINT, new_epoch BIGINT, last_epoch_start BIGINT, new_epoch_start BIGINT)
 LANGUAGE plpgsql
 AS $$
 BEGIN
     EXECUTE format('ALTER TABLE %I DETACH PARTITION %I_partition_%s', table_name, table_name, last_epoch);
-    EXECUTE format('ALTER TABLE %I ATTACH PARTITION %I_partition_%s FOR VALUES FROM (%L) TO (%L)', table_name, table_name, last_epoch, last_epoch_start_cp, new_epoch_start_cp);
-    EXECUTE format('CREATE TABLE IF NOT EXISTS %I_partition_%s PARTITION OF %I FOR VALUES FROM (%L) TO (MAXVALUE)', table_name, new_epoch, table_name, new_epoch_start_cp);
+    EXECUTE format('ALTER TABLE %I ATTACH PARTITION %I_partition_%s FOR VALUES FROM (%L) TO (%L)', table_name, table_name, last_epoch, last_epoch_start, new_epoch_start);
+    EXECUTE format('CREATE TABLE IF NOT EXISTS %I_partition_%s PARTITION OF %I FOR VALUES FROM (%L) TO (MAXVALUE)', table_name, new_epoch, table_name, new_epoch_start);
 END;
 $$;
 

crates/sui-indexer/migrations/pg/2024-05-05-155158_obj_indices/down.sql

@@ -0,0 +1 @@
+DROP TABLE IF EXISTS objects_version;

crates/sui-indexer/migrations/pg/2024-05-05-155158_obj_indices/up.sql

@@ -0,0 +1,31 @@
+-- Indexing table mapping an object's ID and version to its checkpoint
+-- sequence number, partitioned by the first byte of its Object ID.
+CREATE TABLE objects_version (
+    object_id           bytea         NOT NULL,
+    object_version      bigint        NOT NULL,
+    cp_sequence_number  bigint        NOT NULL,
+    PRIMARY KEY (object_id, object_version)
+) PARTITION BY RANGE (object_id);
+
+-- Create a partition for each first byte value.
+DO $$
+DECLARE
+    lo text;
+    hi text;
+BEGIN
+    FOR i IN 0..254 LOOP
+        lo := LPAD(TO_HEX(i), 2, '0');
+        hi := LPAD(TO_HEX(i + 1), 2, '0');
+        EXECUTE FORMAT($F$
+            CREATE TABLE objects_version_%1$s PARTITION OF objects_version FOR VALUES
+            FROM (E'\\x%1$s00000000000000000000000000000000000000000000000000000000000000')
+            TO   (E'\\x%2$s00000000000000000000000000000000000000000000000000000000000000');
+        $F$, lo, hi);
+    END LOOP;
+END;
+$$ LANGUAGE plpgsql;
+
+-- Special case for the last partition, because of the upper bound.
+CREATE TABLE objects_version_ff PARTITION OF objects_version FOR VALUES
+FROM (E'\\xff00000000000000000000000000000000000000000000000000000000000000')
+TO   (MAXVALUE);

crates/sui-indexer/migrations/pg/2024-06-14-045801_event_indices/down.sql

@@ -0,0 +1,7 @@
+DROP TABLE IF EXISTS event_emit_package;
+DROP TABLE IF EXISTS event_emit_module;
+DROP TABLE IF EXISTS event_struct_package;
+DROP TABLE IF EXISTS event_struct_module;
+DROP TABLE IF EXISTS event_struct_name;
+DROP TABLE IF EXISTS event_struct_type;
+DROP TABLE IF EXISTS event_senders;