Fix startup on brand new chain and catchup after long break.

src/tasks/index.rs

@@ -108,20 +108,33 @@ pub async fn backfill_to_block(
     rpc: &Provider<Http>,
     latest_block: Block<H256>,
 ) -> eyre::Result<()> {
+    // Get the first block from the stream and backfill any missing blocks
+    let latest_block_number = latest_block
+        .number
+        .context("Missing block number")?
+        .as_u64();
+
     let next_block_number: u64 = if let Some(latest_db_block_number) =
         app.db.get_latest_block_number(chain_id).await?
     {
         latest_db_block_number + 1
     } else {
-        tracing::info!(chain_id, "No latest block");
-        0
-    };
+        tracing::info!(
+            chain_id,
+            "No latest block in database. Will choose best candidate."
+        );
 
-    // Get the first block from the stream and backfill any missing blocks
-    let latest_block_number = latest_block
-        .number
-        .context("Missing block number")?
-        .as_u64();
+        // Because we do not store all the blocks (we clean up older blocks) there is no need
+        // to scan ALL the blocks. Especially as this may take a lot of time... We are trying
+        // here to move back in time "enough" to get some estimates later.
+        let jump_back_blocks = 60u64;
+
+        if latest_block_number > jump_back_blocks {
+            latest_block_number - jump_back_blocks
+        } else {
+            0
+        }
+    };
 
     tracing::info!(
         latest_block_number,

src/tasks/prune.rs

@@ -12,12 +12,12 @@ const fn minutes(seconds: i64) -> i64 {
     seconds * 60
 }
 
-const fn hours(seconds: i64) -> i64 {
-    minutes(seconds) * 60
+const fn hours(hours: i64) -> i64 {
+    minutes(hours) * 60
 }
 
-const fn days(seconds: i64) -> i64 {
-    hours(seconds) * 24
+const fn days(days: i64) -> i64 {
+    hours(days) * 24
 }
 
 // TODO: This should be a per network setting