fix: cleanup indexer log decoding & RPC polling

crates/registries/src/world_id.rs

@@ -12,7 +12,7 @@ use alloy::{
 sol!(
     /// The registry of World IDs. Each World ID is represented as a leaf in the Merkle tree.
     #[allow(clippy::too_many_arguments)]
-    #[sol(rpc, ignore_unlinked)]
+    #[sol(rpc, ignore_unlinked, extra_derives(Debug, PartialEq, Eq))]
     WorldIdRegistry,
     "abi/WorldIDRegistryAbi.json"
 );

local-setup.sh

@@ -150,7 +150,7 @@ start_node() {
 run_indexer_and_gateway() {
     # remove the tree_cache_file as we have a new DB everytime we run local_setup
     rm -f /tmp/tree.mmap
-    REGISTRY_ADDRESS=$world_id_registry RPC_URL=http://localhost:8545 WS_URL=ws://localhost:8545 DATABASE_URL=postgres://postgres:postgres@localhost:5432/postgres TREE_CACHE_FILE=/tmp/tree.mmap cargo run --release -p world-id-indexer -- --http --indexer > logs/world-id-indexer.log 2>&1 &
+    REGISTRY_ADDRESS=$world_id_registry RPC_URL=http://localhost:8545 DATABASE_URL=postgres://postgres:postgres@localhost:5432/postgres TREE_CACHE_FILE=/tmp/tree.mmap cargo run --release -p world-id-indexer -- --http --indexer > logs/world-id-indexer.log 2>&1 &
     indexer_pid=$!
     echo "started indexer with PID $indexer_pid"
     wait_for_health 8080 "world-id-indexer" 300

run-indexer-stage.sh

@@ -11,7 +11,6 @@ REGISTRY_ADDRESS="${REGISTRY_ADDRESS:-0x969947cFED008bFb5e3F32a25A1A2CDdf64d46fe
 BATCH_SIZE="${BATCH_SIZE:-512}"
 START_BLOCK="${START_BLOCK:-22827118}"
 RPC_URL="${RPC_URL:-https://worldchain-mainnet.g.alchemy.com/public}"
-WS_URL="${WS_URL:-wss://worldchain-mainnet.g.alchemy.com/public}"
 DATABASE_URL="${DATABASE_URL:-postgres://postgres:postgres@localhost:5433/indexer_stage}"
 TREE_CACHE_FILE="${TREE_CACHE_FILE:-$ROOT_DIR/tree-cache-stage.mmap}"
 DB_NAME="${DB_NAME:-indexer_stage}"
@@ -69,7 +68,6 @@ env \
   RUN_MODE="$RUN_MODE" \
   DATABASE_URL="$DATABASE_URL" \
   RPC_URL="$RPC_URL" \
-  WS_URL="$WS_URL" \
   REGISTRY_ADDRESS="$REGISTRY_ADDRESS" \
   START_BLOCK="$START_BLOCK" \
   BATCH_SIZE="$BATCH_SIZE" \

services/indexer/.env.example

@@ -5,10 +5,13 @@ DATABASE_URL=postgres://postgres:postgres@localhost:5432/postgres
 
 # Indexer Config
 RPC_URL=http://localhost:8545
-WS_URL=ws://localhost:8545
 REGISTRY_ADDRESS=0x0000000000000000000000000000000000000000
 START_BLOCK=0
 BATCH_SIZE=64
+# Blocks to stay behind the chain head when polling (avoids ingesting reorg-prone unconfirmed blocks)
+CONFIRMATIONS=8
+# Seconds to wait between RPC poll cycles once caught up
+POLL_INTERVAL_SECS=2
 
 # HTTP Config
 HTTP_ADDR=0.0.0.0:8080

services/indexer/Cargo.toml

@@ -7,7 +7,8 @@ publish = false
 
 [dependencies]
 world-id-services-common = { workspace = true }
-tokio = { workspace = true, features = ["rt-multi-thread", "macros"] }
+tokio = { workspace = true, features = ["rt-multi-thread", "macros", "sync"] }
+backon = { workspace = true }
 tracing = { workspace = true }
 tracing-subscriber = { workspace = true, features = ["json"] }
 telemetry-batteries = { workspace = true }

services/indexer/README.md

@@ -14,7 +14,7 @@ The indexer supports three run modes, selected via the `RUN_MODE` environment va
 
 ### Event Processing
 
-The indexer connects to the chain via HTTP + WebSocket RPC. On startup it backfills from the last indexed block (or `START_BLOCK` if the DB is empty), then transitions to live WebSocket streaming.
+The indexer connects to the chain via HTTP RPC and polls for events. On startup it backfills from the last indexed block (or `START_BLOCK` if the DB is empty), then keeps polling for newly confirmed logs. Only logs at or below the confirmed head (`chain_head - CONFIRMATIONS`) are ingested, so short-lived reorgs of unconfirmed blocks are never committed.
 
 Events are processed through `EventsCommitter`, which:
 
@@ -65,12 +65,13 @@ This restart-on-reorg pattern — detect, rollback state, exit cleanly, re-initi
 |---|---|---|
 | `DATABASE_URL` | required | PostgreSQL connection string |
 | `RPC_URL` | required | HTTP RPC endpoint |
-| `WS_URL` | required | WebSocket RPC endpoint |
 | `REGISTRY_ADDRESS` | required | `AccountRegistry` contract address |
 | `TREE_CACHE_FILE` | required | Path to mmap-backed tree cache file |
 | `RUN_MODE` | `both` | `both`, `indexer`, or `http` |
 | `START_BLOCK` | `0` | Block to start indexing from if DB is empty |
 | `BATCH_SIZE` | `64` | Blocks per RPC batch during backfill |
+| `CONFIRMATIONS` | `8` | Blocks to stay behind the chain head when polling |
+| `POLL_INTERVAL_SECS` | `2` | Seconds between RPC poll cycles once caught up |
 | `TREE_DEPTH` | `30` | Merkle tree depth |
 | `TREE_MAX_BLOCK_AGE` | `1000` | Blocks of per-leaf history kept for rollback |
 | `HTTP_ADDR` | `0.0.0.0:8080` | Address for the HTTP server |