[release/5.x] Cherry pick: Logging nulled snapshots (#7298) (#7302)

Co-authored-by: Amaury Chamayou <amchamay@microsoft.com>
Co-authored-by: Amaury Chamayou <amaury@xargs.fr>

Author: cjen1-msft

CHANGELOG.md

@@ -11,6 +11,7 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 
 ### Added
 
+- Better logging of invalid snapshots (#7302)
 - Logging of snapshot digests
 
 ## [5.0.22]

src/node/snapshot_serdes.h

@@ -50,6 +50,11 @@ namespace ccf
     auto store_snapshot_size =
       sizeof(ccf::kv::SerialisedEntryHeader) + tx_hdr.size;
 
+    if (tx_hdr.size == 0)
+    {
+      throw std::logic_error("Snapshot transaction size should not be zero");
+    }
+
     auto receipt_data = data + store_snapshot_size;
     auto receipt_size = size - store_snapshot_size;
 
@@ -58,6 +63,20 @@ namespace ccf
       throw std::logic_error("No receipt included in snapshot");
     }
 
+    LOG_INFO_FMT("Deserialising snapshot receipt (size: {}).", receipt_size);
+    constexpr size_t max_printed_size = 1024;
+    if (receipt_size > max_printed_size)
+    {
+      LOG_INFO_FMT(
+        "Receipt size ({}) exceeds max printed size ({}), only printing "
+        "first {} bytes",
+        receipt_size,
+        max_printed_size,
+        max_printed_size);
+    }
+    auto printed_size = std::min<size_t>(receipt_size, max_printed_size);
+    LOG_INFO_FMT("{}", ds::to_hex(receipt_data, receipt_data + printed_size));
+
     auto j = nlohmann::json::parse(receipt_data, receipt_data + receipt_size);
     auto receipt_p = j.get<ReceiptPtr>();
     auto receipt = std::dynamic_pointer_cast<ccf::ProofReceipt>(receipt_p);

tests/e2e_operations.py

@@ -288,6 +288,40 @@ def test_empty_snapshot(network, args):
                 )
 
 
+def test_nulled_snapshot(network, args):
+
+    with tempfile.TemporaryDirectory() as snapshots_dir:
+        LOG.debug(f"Using {snapshots_dir} as snapshots directory")
+
+        snapshot_name = "snapshot_1000_1500.committed"
+
+        with open(
+            os.path.join(snapshots_dir, snapshot_name), "wb+"
+        ) as temp_empty_snapshot:
+
+            LOG.debug(f"Created empty snapshot {temp_empty_snapshot.name}")
+            temp_empty_snapshot.write(b"\x00" * 64)
+
+        LOG.info(
+            "Attempt to join a node using the corrupted snapshot copy (should fail)"
+        )
+        new_node = network.create_node("local://localhost")
+        failed = False
+        try:
+            network.join_node(
+                new_node,
+                args.package,
+                args,
+                snapshots_dir=snapshots_dir,
+            )
+        except Exception as e:
+            failed = True
+            LOG.info(f"Node failed to join as expected: {e}")
+
+        # (Existing assertion logic retained)
+        assert failed, "Node should not have joined successfully"
+
+
 def split_all_ledger_files_in_dir(input_dir, output_dir):
     # A ledger file can only be split at a seqno that contains a signature
     # (so that all files end on a signature that verifies their integrity).
@@ -376,6 +410,7 @@ def run_file_operations(args):
                 test_forced_snapshot(network, args)
                 test_large_snapshot(network, args)
                 test_empty_snapshot(network, args)
+                test_nulled_snapshot(network, args)
 
                 primary, _ = network.find_primary()
                 # Scoped transactions are not handled by historical range queries