From 7e4206054104a0128c535c31d97d4e205dd8d252 Mon Sep 17 00:00:00 2001
From: Prathamesh Musale <prathamesh.musale0@gmail.com>
Date: Thu, 6 Jun 2024 04:33:25 +0000
Subject: [PATCH] Add config for geth dir path in snapshot archive (#2)

Part of [Create a stack to run a BSC node](https://www.notion.so/Create-a-stack-to-run-a-BSC-node-5740b35b02974661a0237dcf57ab8150)

Reviewed-on: https://git.vdb.to/cerc-io/bsc-stack/pulls/2
Co-authored-by: Prathamesh Musale <prathamesh.musale0@gmail.com>
Co-committed-by: Prathamesh Musale <prathamesh.musale0@gmail.com>
---
 .../compose/docker-compose-bsc-node.yml       |  1 +
 .../config/bsc-node/docker-entrypoint.sh      | 21 +++++++++++--------
 stack-orchestrator/stacks/bsc-node/README.md  |  4 ++++
 3 files changed, 17 insertions(+), 9 deletions(-)

diff --git a/stack-orchestrator/compose/docker-compose-bsc-node.yml b/stack-orchestrator/compose/docker-compose-bsc-node.yml
index 9789a63..b797e7b 100644
--- a/stack-orchestrator/compose/docker-compose-bsc-node.yml
+++ b/stack-orchestrator/compose/docker-compose-bsc-node.yml
@@ -5,6 +5,7 @@ services:
     environment:
       CERC_BSC_NETWORK: ${CERC_BSC_NETWORK:-mainnet}
       CERC_USE_SNAPSHOT: ${CERC_USE_SNAPSHOT:-false}
+      CERC_SNAPSHOT_GETH_PATH: ${CERC_SNAPSHOT_GETH_PATH}
       CERC_FAST_NODE: ${CERC_FAST_NODE:-false}
       CERC_GETH_GCMODE: ${CERC_GETH_GCMODE:-full}
       CERC_GETH_VERBOSITY: ${CERC_GETH_VERBOSITY:-3}
diff --git a/stack-orchestrator/config/bsc-node/docker-entrypoint.sh b/stack-orchestrator/config/bsc-node/docker-entrypoint.sh
index 056a1d0..0d155c3 100755
--- a/stack-orchestrator/config/bsc-node/docker-entrypoint.sh
+++ b/stack-orchestrator/config/bsc-node/docker-entrypoint.sh
@@ -28,20 +28,23 @@ ADDITIONAL_FLAGS=""
 if [ ! -d "$geth_dir" ] && [ "${CERC_USE_SNAPSHOT}" = "true" ] && [ -n "$snapshot_file" ]; then
   echo "Using snapshot $snapshot_file"
 
+  if [ -z "$CERC_SNAPSHOT_GETH_PATH" ]; then
+    echo CERC_SNAPSHOT_GETH_PATH not set
+    exit 1
+  else
+    # Remove any trailing / from the given path
+    archive_geth_path=$(echo "$CERC_SNAPSHOT_GETH_PATH" | sed 's:/*$::')
+
+    # Determine strip-components count from the geth dir path
+    to_strip=$(echo "$archive_geth_path" | grep -o '/' | wc -l)
+  fi
+
   case $snapshot_file in
   *.lz4)
-    # Determine where geth dir is located
-    archive_geth_path=$(lz4 -c -d $snapshot_file | tar -tf - | grep -m1 -E 'geth/?$')
-
-    # Determine strip-components count from the path
-    to_strip=$(($(echo "$archive_geth_path" | grep -o '/' | wc -l) - 1))
-
+    # Extract geth data directly to destination directory
     lz4 -c -d $snapshot_file | tar -C ${DATA_DIR}/ -xvf - --strip-components=$to_strip $archive_geth_path
     ;;
   *.zst)
-    archive_geth_path=$(zstd -c -d $snapshot_file | tar -tf - | grep -m1 -E 'geth/?$')
-    to_strip=$(($(echo "$archive_geth_path" | grep -o '/' | wc -l) - 1))
-
     zstd -c -d $snapshot_file | tar -C ${DATA_DIR}/ -xvf - --strip-components=$to_strip $archive_geth_path
     ;;
   *)
diff --git a/stack-orchestrator/stacks/bsc-node/README.md b/stack-orchestrator/stacks/bsc-node/README.md
index 9853f6f..13a5f26 100644
--- a/stack-orchestrator/stacks/bsc-node/README.md
+++ b/stack-orchestrator/stacks/bsc-node/README.md
@@ -64,6 +64,10 @@ Instructions for running a BSC node
   # Requires snapshot file with extension .tar.lz4 or .tar.zst to be placed in the snapshot directory
   CERC_USE_SNAPSHOT=
 
+  # Path where geth dir is located in the snapshot archive
+  # Required when using a snapshot; example: "server/data-seed/geth"
+  CERC_SNAPSHOT_GETH_PATH=
+
   # Whether to run as a fast node (true | false) (default: false)
   # Fast node does not generate trie data when syncing
   CERC_FAST_NODE=