Limit the max open files of state sync cache db

Fixes #1479

Author: cburgdorf

eth/db/backends/level.py

@@ -28,7 +28,9 @@ class LevelDB(BaseAtomicDB):
     logger = logging.getLogger("eth.db.backends.LevelDB")
 
     # Creates db as a class variable to avoid level db lock error
-    def __init__(self, db_path: Path = None) -> None:
+    def __init__(self,
+                 db_path: Path=None,
+                 max_open_files: int=None) -> None:
         if not db_path:
             raise TypeError("Please specifiy a valid path for your database.")
         try:
@@ -39,7 +41,12 @@ def __init__(self, db_path: Path = None) -> None:
                 "LevelDB requires the plyvel library which is not available for import."
             )
         self.db_path = db_path
-        self.db = plyvel.DB(str(db_path), create_if_missing=True, error_if_exists=False)
+        self.db = plyvel.DB(
+            str(db_path),
+            create_if_missing=True,
+            error_if_exists=False,
+            max_open_files=max_open_files
+        )
 
     def __getitem__(self, key: bytes) -> bytes:
         v = self.db.get(key)

trinity/sync/full/state.py

@@ -62,6 +62,7 @@
     HexaryTrieSync,
     SyncRequest,
 )
+from trinity.utils.os import get_open_fd_limit
 from trinity.utils.timer import Timer
 
 
@@ -85,8 +86,17 @@ def __init__(self,
         # We use a LevelDB instance for the nodes cache because a full state download, if run
         # uninterrupted will visit more than 180M nodes, making an in-memory cache unfeasible.
         self._nodes_cache_dir = tempfile.TemporaryDirectory(prefix="pyevm-state-sync-cache")
+
+        # Allow the LevelDB instance to consume half of the entire file descriptor limit that
+        # the OS permits. Let the other half be reserved for other db access, networking etc.
+        max_open_files = get_open_fd_limit() // 2
+
         self.scheduler = StateSync(
-            root_hash, account_db, LevelDB(cast(Path, self._nodes_cache_dir.name)), self.logger)
+            root_hash,
+            account_db,
+            LevelDB(Path(self._nodes_cache_dir.name), max_open_files),
+            self.logger
+        )
         self.request_tracker = TrieNodeRequestTracker(self._reply_timeout, self.logger)
         self._peer_missing_nodes: Dict[ETHPeer, Set[Hash32]] = collections.defaultdict(set)
 

trinity/utils/os.py

@@ -0,0 +1,9 @@
+import resource
+
+
+def get_open_fd_limit() -> int:
+    """
+    Return the OS soft limit of open file descriptors per process.
+    """
+    soft_limit, hard_limit = resource.getrlimit(resource.RLIMIT_NOFILE)
+    return soft_limit