Merge pull request #2111 from IntersectMBO/fix/lsm-snapshot-doubled-path

Fix LSM snapshot restoration by resolving double path nesting

Author: kderme

cardano-db-sync/src/Cardano/DbSync/Ledger/State.hs

@@ -118,7 +118,7 @@ import Ouroboros.Network.Block (HeaderHash, pointSlot)
 import System.FS.API (SomeHasFS (..), mkFsPath)
 import System.FS.API.Types (MountPoint (..))
 import System.FS.IO (ioHasFS)
-import System.FilePath (splitDirectories, (</>))
+import System.FilePath ((</>))
 import System.Mem (performMajorGC)
 import System.Random (genWord64, newStdGen)
 
@@ -227,7 +227,13 @@ mkHasLedgerEnv trce protoInfo dir nw maxLovelaceSupply systemStart syncOptions b
     LedgerBackendLSM mPath -> do
       let lsmPath = fromMaybe (unLedgerStateDir dir </> "lsm") mPath
       salt <- fst . genWord64 <$> newStdGen
-      let args = LSM.LSMArgs (mkFsPath $ splitDirectories lsmPath) salt (LSM.stdMkBlockIOFS lsmPath)
+      -- The HasBlockIO is rooted at lsmPath, so the session's FsPath inside it
+      -- must be the empty path. Passing the full lsmPath here causes the LSM
+      -- session to be created at <lsmPath>/<lsmPath>, which breaks snapshot
+      -- bundling in scripts/postgresql-setup.sh (it falls back to the InMemory
+      -- branch and ships an incomplete tarball; on restore db-sync then can't
+      -- find the snapshot and replays from genesis).
+      let args = LSM.LSMArgs (mkFsPath []) salt (LSM.stdMkBlockIOFS lsmPath)
       res <-
         runWithTempRegistry $
           (,())

scripts/postgresql-setup.sh

@@ -176,6 +176,15 @@ function create_snapshot {
 	      # The active/ directory is not needed — it gets cleared on session restore.
 	      tar czf "${tmp_dir}/${ledger_dir_name}.tar.gz" \
 	        -C "${state_dir}" "${ledger_dir_name}" "lsm/snapshots/${ledger_dir_name}" "lsm/metadata"
+	    elif [ -d "${state_dir}/lsm" ]; then
+	      # State directory contains an lsm/ subtree but the expected snapshot path is missing.
+	      # Refuse to silently fall back to the InMemory branch — the resulting tarball would
+	      # exclude the LSM tree files and restoration would replay from genesis.
+	      echo "ERROR: ${state_dir}/lsm exists but no LSM snapshot found at ${lsm_snap_dir}." >&2
+	      echo "       The LSM session may be using a non-standard layout. Aborting to avoid" >&2
+	      echo "       producing an incomplete snapshot." >&2
+	      rm "${recursive}" "${force}" "${tmp_dir}"
+	      exit 1
 	    else
 	      echo "Detected InMemory backend snapshot at slot ${ledger_dir_name}"
 	      # InMemory backend: just the ledger snapshot directory