3137: Bump Python SDK version requirement. Tweak Stats() class, don't keep two
[arvados.git] / services / fuse / bin / arv-mount
index 5b161fd2be57345dc52b1b120dc5079ecca3f28e..44bc698299975230792de112fa2644e6b458c7b8 100755 (executable)
@@ -11,9 +11,71 @@ import time
 
 import arvados.commands._util as arv_cmd
 from arvados_fuse import *
+from arvados.safeapi import ThreadSafeApiCache
+import arvados.keep
 
 logger = logging.getLogger('arvados.arv-mount')
 
+class Stat(object):
+    def __init__(self, prefix, interval,
+                 egr_name, ing_name,
+                 egr_func, ing_func):
+        self.prefix = prefix
+        self.interval = interval
+        self.egr_name = egr_name
+        self.ing_name = ing_name
+        self.egress = egr_func
+        self.ingress = ing_func
+        self.egr_prev = self.egress()
+        self.ing_prev = self.ingress()
+
+    def update(self):
+        egr = self.egress()
+        ing = self.ingress()
+
+        delta = " -- interval %.4f seconds %d %s %d %s" % (self.interval,
+                                                           egr - self.egr_prev,
+                                                           self.egr_name,
+                                                           ing - self.ing_prev,
+                                                           self.ing_name)
+
+        sys.stderr.write("crunchstat: %s %d %s %d %s%s\n" % (self.prefix,
+                                                             egr,
+                                                             self.egr_name,
+                                                             ing,
+                                                             self.ing_name,
+                                                             delta))
+
+        self.egr_prev = egr
+        self.ing_prev = ing
+
+
+def statlogger(interval, keep, ops):
+    calls = Stat("keepcalls", interval, "put", "get",
+                 keep.put_counter.get,
+                 keep.get_counter.get)
+    net = Stat("net:keep0", interval, "tx", "rx",
+               keep.upload_counter.get,
+               keep.download_counter.get)
+    cache = Stat("keepcache", interval, "hit", "miss",
+               keep.hits_counter.get,
+               keep.misses_counter.get)
+    fuseops = Stat("fuseops", interval,"write", "read",
+                   ops.write_ops_counter.get,
+                   ops.read_ops_counter.get)
+    blk = Stat("blkio:0:0", interval, "write", "read",
+               ops.write_counter.get,
+               ops.read_counter.get)
+
+    while True:
+        time.sleep(interval)
+        calls.update()
+        net.update()
+        cache.update()
+        fuseops.update()
+        blk.update()
+
+
 if __name__ == '__main__':
     # Handle command line parameters
     parser = argparse.ArgumentParser(
@@ -37,6 +99,8 @@ with "--".
                             help="""Mount subdirectories listed by tag.""")
     mount_mode.add_argument('--by-id', action='store_true',
                             help="""Mount subdirectories listed by portable data hash or uuid.""")
+    mount_mode.add_argument('--by-pdh', action='store_true',
+                            help="""Mount subdirectories listed by portable data hash.""")
     mount_mode.add_argument('--project', type=str, help="""Mount a specific project.""")
     mount_mode.add_argument('--collection', type=str, help="""Mount only the specified collection.""")
 
@@ -44,6 +108,15 @@ with "--".
     parser.add_argument('--logfile', help="""Write debug logs and errors to the specified file (default stderr).""")
     parser.add_argument('--foreground', action='store_true', help="""Run in foreground (default is to daemonize unless --exec specified)""", default=False)
     parser.add_argument('--encoding', type=str, help="Character encoding to use for filesystem, default is utf-8 (see Python codec registry for list of available encodings)", default="utf-8")
+
+    parser.add_argument('--file-cache', type=int, help="File data cache size, in bytes (default 256MiB)", default=256*1024*1024)
+    parser.add_argument('--directory-cache', type=int, help="Directory data cache size, in bytes (default 128MiB)", default=128*1024*1024)
+
+    parser.add_argument('--read-only', action='store_false', help="Mount will be read only (default)", dest="enable_write", default=False)
+    parser.add_argument('--read-write', action='store_true', help="Mount will be read-write", dest="enable_write", default=False)
+
+    parser.add_argument('--crunchstat-interval', type=float, help="Write stats to stderr every N seconds (default disabled)", default=0)
+
     parser.add_argument('--exec', type=str, nargs=argparse.REMAINDER,
                         dest="exec_args", metavar=('command', 'args', '...', '--'),
                         help="""Mount, run a command, then unmount and exit""")
@@ -78,18 +151,31 @@ with "--".
         arvados.logger.setLevel(logging.DEBUG)
         logger.debug("arv-mount debugging enabled")
 
+    logger.info("enable write is %s", args.enable_write)
+
     try:
         # Create the request handler
-        operations = Operations(os.getuid(), os.getgid(), args.encoding)
-        api = SafeApi(arvados.config)
+        operations = Operations(os.getuid(),
+                                os.getgid(),
+                                encoding=args.encoding,
+                                inode_cache=InodeCache(cap=args.directory_cache),
+                                enable_write=args.enable_write)
+        api = ThreadSafeApiCache(apiconfig=arvados.config.settings(),
+                                 keep_params={"block_cache": arvados.keep.KeepBlockCache(args.file_cache)})
+
+        if args.crunchstat_interval:
+            statsthread = threading.Thread(target=statlogger, args=(args.crunchstat_interval, api.keep, operations))
+            statsthread.daemon = True
+            statsthread.start()
 
         usr = api.users().current().execute(num_retries=args.retries)
         now = time.time()
         dir_class = None
         dir_args = [llfuse.ROOT_INODE, operations.inodes, api, args.retries]
-        if args.by_id:
+        if args.by_id or args.by_pdh:
             # Set up the request handler with the 'magic directory' at the root
             dir_class = MagicDirectory
+            dir_args.append(args.by_pdh)
         elif args.by_tag:
             dir_class = TagsDirectory
         elif args.shared:
@@ -98,6 +184,7 @@ with "--".
         elif args.home:
             dir_class = ProjectDirectory
             dir_args.append(usr)
+            dir_args.append(True)
         elif args.collection is not None:
             # Set up the request handler with the collection at the root
             dir_class = CollectionDirectory
@@ -110,13 +197,15 @@ with "--".
         if dir_class is not None:
             operations.inodes.add_entry(dir_class(*dir_args))
         else:
-            e = operations.inodes.add_entry(Directory(llfuse.ROOT_INODE))
+            e = operations.inodes.add_entry(Directory(llfuse.ROOT_INODE, operations.inodes))
             dir_args[0] = e.inode
 
             e._entries['by_id'] = operations.inodes.add_entry(MagicDirectory(*dir_args))
+
             e._entries['by_tag'] = operations.inodes.add_entry(TagsDirectory(*dir_args))
 
             dir_args.append(usr)
+            dir_args.append(True)
             e._entries['home'] = operations.inodes.add_entry(ProjectDirectory(*dir_args))
             e._entries['shared'] = operations.inodes.add_entry(SharedDirectory(*dir_args))
 
@@ -144,10 +233,17 @@ From here, the following directories are available:
     opts = [optname for optname in ['allow_other', 'debug']
             if getattr(args, optname)]
 
+    # Increase default read/write size from 4KiB to 128KiB
+    opts += ["big_writes", "max_read=131072"]
+
     if args.exec_args:
         # Initialize the fuse connection
         llfuse.init(operations, args.mountpoint, opts)
 
+        # Subscribe to change events from API server
+        if not args.by_pdh:
+            operations.listen_for_events(api)
+
         t = threading.Thread(None, lambda: llfuse.main())
         t.start()
 
@@ -179,12 +275,19 @@ From here, the following directories are available:
                 pass
         finally:
             subprocess.call(["fusermount", "-u", "-z", args.mountpoint])
+            operations.destroy()
 
         exit(rc)
     else:
         try:
             llfuse.init(operations, args.mountpoint, opts)
+
+            # Subscribe to change events from API server
+            operations.listen_for_events(api)
+
             llfuse.main()
         except Exception as e:
             logger.exception('arv-mount: exception during mount')
             exit(getattr(e, 'errno', 1))
+        finally:
+            operations.destroy()