fix: fixed nayduck test state_sync_fail.py for nightly build (#9320)

In #9274 I introduced simple nightshade V2 layout and added it to the nightly build. This broke the nayduck test state_sync_fail.py. Here is the fix for it. 

The test performs resharding and then checks some postconditions. It broke because it attempted to reshard from V0 shard layout to V2 shard layout. This doesn't work because ShardLayout contains shard split map that only makes sense when resharding from a shard layout version to the immediate next. 

The fix is to check what is the protocol version supported in the binary and depending on it reshard from V0 to V1 or from V1 to V2.

pytest/lib/cluster.py

@@ -946,3 +946,23 @@ def load_config():
     else:
         logger.info(f"Use default config {config}")
     return config
+
+
+# Returns the protocol version of the binary.
+def get_binary_protocol_version(config) -> typing.Optional[int]:
+    binary_name = config.get('binary_name', 'neard')
+    near_root = config.get('near_root')
+    binary_path = os.path.join(near_root, binary_name)
+
+    # Get the protocol version of the binary
+    # The --version output looks like this:
+    # neard (release trunk) (build 1.1.0-3884-ge93793a61-modified) (rustc 1.71.0) (protocol 137) (db 37)
+    out = subprocess.check_output([binary_path, "--version"], text=True)
+    out = out.replace('(', '')
+    out = out.replace(')', '')
+    tokens = out.split()
+    n = len(tokens)
+    for i in range(n):
+        if tokens[i] == "protocol" and i + 1 < n:
+            return int(tokens[i + 1])
+    return None

pytest/tests/sanity/state_sync_fail.py

@@ -1,29 +1,124 @@
 #!/usr/bin/env python3
-# Spins up a node, wait until sharding is upgrade
-# and spins up another node
-# check that the node can't be started because it cannot state sync to the epoch
-# after the sharding upgrade
+
+# Spins up a node, waits until sharding is upgraded and spins up another node.
+# Check that the node can't be started because it cannot state sync to the epoch
+# after the sharding upgrade.
+
+# Depending on the version of the binary (default or nightly) it will perform
+# resharding from V0 (1 shard) to V1 (4 shards) or from V1 (4 shards) to V2 (5
+# shards).
 
 import sys, time
 import pathlib
 
 sys.path.append(str(pathlib.Path(__file__).resolve().parents[2] / 'lib'))
 
-from cluster import init_cluster, spin_up_node, load_config
+from cluster import init_cluster, spin_up_node, load_config, get_binary_protocol_version
 from configured_logger import logger
 import requests
 import utils
 
 EPOCH_LENGTH = 10
 START_AT_BLOCK = int(EPOCH_LENGTH * 2.5)
 
+V1_PROTOCOL_VERSION = 48
+V2_PROTOCOL_VERSION = 135
+
+V0_SHARD_LAYOUT = {"V0": {"num_shards": 1, "version": 0}}
+V1_SHARD_LAYOUT = {
+    "V1": {
+        "boundary_accounts": [
+            "aurora", "aurora-0", "kkuuue2akv_1630967379.near"
+        ],
+        "shards_split_map": [[0, 1, 2, 3]],
+        "to_parent_shard_map": [0, 0, 0, 0],
+        "version": 1
+    }
+}
+
+
+def append_shard_layout_config_changes(
+    binary_protocol_version,
+    genesis_config_changes,
+):
+    if binary_protocol_version >= V2_PROTOCOL_VERSION:
+        logger.info("Testing migration from V1 to V2.")
+        # Set the initial protocol version to a version just before V2.
+        genesis_config_changes.append([
+            "protocol_version",
+            V2_PROTOCOL_VERSION - 1,
+        ])
+        genesis_config_changes.append([
+            "shard_layout",
+            V1_SHARD_LAYOUT,
+        ])
+        genesis_config_changes.append([
+            "num_block_producer_seats_per_shard",
+            [1, 1, 1, 1],
+        ])
+        genesis_config_changes.append([
+            "avg_hidden_validator_seats_per_shard",
+            [0, 0, 0, 0],
+        ])
+        print(genesis_config_changes)
+        return
+
+    if binary_protocol_version >= V1_PROTOCOL_VERSION:
+        logger.info("Testing migration from V0 to V1.")
+        # Set the initial protocol version to a version just before V1.
+        genesis_config_changes.append([
+            "protocol_version",
+            V1_PROTOCOL_VERSION - 1,
+        ])
+        genesis_config_changes.append([
+            "shard_layout",
+            V0_SHARD_LAYOUT,
+        ])
+        genesis_config_changes.append([
+            "num_block_producer_seats_per_shard",
+            [100],
+        ])
+        genesis_config_changes.append([
+            "avg_hidden_validator_seats_per_shard",
+            [0],
+        ])
+        print(genesis_config_changes)
+        return
+
+    assert False
+
+
+def get_genesis_config_changes(binary_protocol_version):
+    genesis_config_changes = [
+        ["min_gas_price", 0],
+        ["max_inflation_rate", [0, 1]],
+        ["epoch_length", EPOCH_LENGTH],
+        ["use_production_config", True],
+        ["block_producer_kickout_threshold", 80],
+    ]
+
+    append_shard_layout_config_changes(
+        binary_protocol_version,
+        genesis_config_changes,
+    )
+
+    print(genesis_config_changes)
+
+    return genesis_config_changes
+
+
 config = load_config()
+
+binary_protocol_version = get_binary_protocol_version(config)
+assert binary_protocol_version is not None
+
 near_root, node_dirs = init_cluster(
-    2, 1, 1, config,
-    [["min_gas_price", 0], ["max_inflation_rate", [0, 1]],
-     ["epoch_length", EPOCH_LENGTH], ["protocol_version", 47],
-     ["use_production_config", True], ["block_producer_kickout_threshold", 80]],
-    {
+    num_nodes=2,
+    num_observers=1,
+    num_shards=4,
+    config=config,
+    genesis_config_changes=get_genesis_config_changes(binary_protocol_version),
+    client_config_changes={
         0: {
             "tracked_shards": [0],
             "state_sync_enabled": True,
@@ -42,7 +137,8 @@
             "state_sync_enabled": True,
             "store.state_snapshot_enabled": True,
         }
-    })
+    },
+)
 
 started = time.time()