def process_file(file_path: Path) -> Optional[PlotInfo]:
if not self._refreshing_enabled:
return None
filename_str = str(file_path)
if self.match_str is not None and self.match_str not in filename_str:
return None
if (
file_path in self.failed_to_open_filenames
and (time.time() - self.failed_to_open_filenames[file_path])
< self.refresh_parameter.retry_invalid_seconds
):
# Try once every `refresh_parameter.retry_invalid_seconds` seconds to open the file
return None
if file_path in self.plots:
return self.plots[file_path]
entry: Optional[Tuple[str, Set[str]]] = self.plot_filename_paths.get(file_path.name)
if entry is not None:
loaded_parent, duplicates = entry
if str(file_path.parent) in duplicates:
log.debug(f"Skip duplicated plot {str(file_path)}")
return None
try:
if not file_path.exists():
return None
prover = DiskProver(str(file_path))
log.debug(f"process_file {str(file_path)}")
expected_size = _expected_plot_size(prover.get_size()) * UI_ACTUAL_SPACE_CONSTANT_FACTOR
stat_info = file_path.stat()
# TODO: consider checking if the file was just written to (which would mean that the file is still
# being copied). A segfault might happen in this edge case.
if prover.get_size() >= 30 and stat_info.st_size < 0.98 * expected_size:
log.warning(
f"Not farming plot {file_path}. Size is {stat_info.st_size / (1024**3)} GiB, but expected"
f" at least: {expected_size / (1024 ** 3)} GiB. We assume the file is being copied."
)
return None
cache_entry = self.cache.get(prover.get_id())
if cache_entry is None:
(
pool_public_key_or_puzzle_hash,
farmer_public_key,
local_master_sk,
) = parse_plot_info(prover.get_memo())
# Only use plots that correct keys associated with them
if farmer_public_key not in self.farmer_public_keys:
log.warning(f"Plot {file_path} has a farmer public key that is not in the farmer's pk list.")
self.no_key_filenames.add(file_path)
if not self.open_no_key_filenames:
return None
pool_public_key: Optional[G1Element] = None
pool_contract_puzzle_hash: Optional[bytes32] = None
if isinstance(pool_public_key_or_puzzle_hash, G1Element):
pool_public_key = pool_public_key_or_puzzle_hash
else:
assert isinstance(pool_public_key_or_puzzle_hash, bytes32)
pool_contract_puzzle_hash = pool_public_key_or_puzzle_hash
if pool_public_key is not None and pool_public_key not in self.pool_public_keys:
log.warning(f"Plot {file_path} has a pool public key that is not in the farmer's pool pk list.")
self.no_key_filenames.add(file_path)
if not self.open_no_key_filenames:
return None
# If a plot is in `no_key_filenames` the keys were missing in earlier refresh cycles. We can remove
# the current plot from that list if its in there since we passed the key checks above.
if file_path in self.no_key_filenames:
self.no_key_filenames.remove(file_path)
local_sk = master_sk_to_local_sk(local_master_sk)
plot_public_key: G1Element = ProofOfSpace.generate_plot_public_key(
local_sk.get_g1(), farmer_public_key, pool_contract_puzzle_hash is not None
)
cache_entry = CacheEntry(pool_public_key, pool_contract_puzzle_hash, plot_public_key)
self.cache.update(prover.get_id(), cache_entry)
with self.plot_filename_paths_lock:
paths: Optional[Tuple[str, Set[str]]] = self.plot_filename_paths.get(file_path.name)
if paths is None:
paths = (str(Path(prover.get_filename()).parent), set())
self.plot_filename_paths[file_path.name] = paths
else:
paths[1].add(str(Path(prover.get_filename()).parent))
log.warning(f"Have multiple copies of the plot {file_path.name} in {[paths[0], *paths[1]]}.")
return None
new_plot_info: PlotInfo = PlotInfo(
prover,
cache_entry.pool_public_key,
cache_entry.pool_contract_puzzle_hash,
cache_entry.plot_public_key,
stat_info.st_size,
stat_info.st_mtime,
)
with counter_lock:
result.loaded.append(new_plot_info)
if file_path in self.failed_to_open_filenames:
del self.failed_to_open_filenames[file_path]
except Exception as e:
tb = traceback.format_exc()
log.error(f"Failed to open file {file_path}. {e} {tb}")
self.failed_to_open_filenames[file_path] = int(time.time())
return None
log.info(f"Found plot {file_path} of size {new_plot_info.prover.get_size()}")
if self.show_memo:
plot_memo: bytes32
if pool_contract_puzzle_hash is None:
plot_memo = stream_plot_info_pk(pool_public_key, farmer_public_key, local_master_sk)
else:
plot_memo = stream_plot_info_ph(pool_contract_puzzle_hash, farmer_public_key, local_master_sk)
plot_memo_str: str = plot_memo.hex()
log.info(f"Memo: {plot_memo_str}")
return new_plot_info
def process_file(directory: Path, filename: Path) -> Tuple[int, Dict]:
new_provers: Dict[Path, PlotInfo] = {}
nonlocal changed
filename_str = str(filename)
if match_str is not None and match_str not in filename_str:
return 0, new_provers
if filename.exists():
if filename in failed_to_open_filenames and (
time.time() - failed_to_open_filenames[filename]) < 1200:
# Try once every 20 minutes to open the file
return 0, new_provers
if filename in provers:
try:
stat_info = filename.stat()
except Exception as e:
log.error(f"Failed to open file {filename}. {e}")
return 0, new_provers
if stat_info.st_mtime == provers[filename].time_modified:
with plot_ids_lock:
if provers[filename].prover.get_id() in plot_ids:
log.warning(
f"Have multiple copies of the plot {filename}, not adding it."
)
return 0, new_provers
plot_ids.add(provers[filename].prover.get_id())
new_provers[filename] = provers[filename]
return stat_info.st_size, new_provers
try:
prover = DiskProver(str(filename))
expected_size = _expected_plot_size(
prover.get_size()) * UI_ACTUAL_SPACE_CONSTANT_FACTOR
stat_info = filename.stat()
# TODO: consider checking if the file was just written to (which would mean that the file is still
# being copied). A segfault might happen in this edge case.
if prover.get_size(
) >= 30 and stat_info.st_size < 0.98 * expected_size:
log.warning(
f"Not farming plot {filename}. Size is {stat_info.st_size / (1024**3)} GiB, but expected"
f" at least: {expected_size / (1024 ** 3)} GiB. We assume the file is being copied."
)
return 0, new_provers
(
pool_public_key_or_puzzle_hash,
farmer_public_key,
local_master_sk,
) = parse_plot_info(prover.get_memo())
# Only use plots that correct keys associated with them
if farmer_public_keys is not None and farmer_public_key not in farmer_public_keys:
log.warning(
f"Plot {filename} has a farmer public key that is not in the farmer's pk list."
)
no_key_filenames.add(filename)
if not open_no_key_filenames:
return 0, new_provers
if isinstance(pool_public_key_or_puzzle_hash, G1Element):
pool_public_key = pool_public_key_or_puzzle_hash
pool_contract_puzzle_hash = None
else:
assert isinstance(pool_public_key_or_puzzle_hash, bytes32)
pool_public_key = None
pool_contract_puzzle_hash = pool_public_key_or_puzzle_hash
if (pool_public_keys is not None
and pool_public_key is not None
and pool_public_key not in pool_public_keys):
log.warning(
f"Plot {filename} has a pool public key that is not in the farmer's pool pk list."
)
no_key_filenames.add(filename)
if not open_no_key_filenames:
return 0, new_provers
stat_info = filename.stat()
local_sk = master_sk_to_local_sk(local_master_sk)
plot_public_key: G1Element = ProofOfSpace.generate_plot_public_key(
local_sk.get_g1(), farmer_public_key)
with plot_ids_lock:
if prover.get_id() in plot_ids:
log.warning(
f"Have multiple copies of the plot {filename}, not adding it."
)
return 0, new_provers
plot_ids.add(prover.get_id())
new_provers[filename] = PlotInfo(
prover,
pool_public_key,
pool_contract_puzzle_hash,
plot_public_key,
stat_info.st_size,
stat_info.st_mtime,
directory,
)
changed = True
except Exception as e:
tb = traceback.format_exc()
log.error(f"Failed to open file {filename}. {e} {tb}")
failed_to_open_filenames[filename] = int(time.time())
return 0, new_provers
log.info(
f"Found plot {filename} of size {new_provers[filename].prover.get_size()}"
)
if show_memo:
plot_memo: bytes32
if pool_contract_puzzle_hash is None:
plot_memo = stream_plot_info_pk(pool_public_key,
farmer_public_key,
local_master_sk)
else:
plot_memo = stream_plot_info_ph(pool_contract_puzzle_hash,
farmer_public_key,
local_master_sk)
plot_memo_str: str = plot_memo.hex()
log.info(f"Memo: {plot_memo_str}")
return stat_info.st_size, new_provers
return 0, new_provers
def process_file(file_path: Path) -> Dict:
new_provers: Dict[Path, PlotInfo] = {}
filename_str = str(file_path)
if self.match_str is not None and self.match_str not in filename_str:
return new_provers
if file_path.exists():
if (
file_path in self.failed_to_open_filenames
and (time.time() - self.failed_to_open_filenames[file_path]) > 1200
):
# Try once every 20 minutes to open the file
return new_provers
if file_path in self.plots:
try:
stat_info = file_path.stat()
except Exception as e:
log.error(f"Failed to open file {file_path}. {e}")
return new_provers
if stat_info.st_mtime == self.plots[file_path].time_modified:
new_provers[file_path] = self.plots[file_path]
return new_provers
entry: Optional[Tuple[str, Set[str]]] = self.plot_filename_paths.get(file_path.name)
if entry is not None:
loaded_parent, duplicates = entry
if str(file_path.parent) in duplicates:
log.debug(f"Skip duplicated plot {str(file_path)}")
return new_provers
try:
with counter_lock:
if result.processed_files >= self.refresh_parameter.batch_size:
result.remaining_files += 1
return new_provers
result.processed_files += 1
prover = DiskProver(str(file_path))
log.debug(f"process_file {str(file_path)}")
expected_size = _expected_plot_size(prover.get_size()) * UI_ACTUAL_SPACE_CONSTANT_FACTOR
stat_info = file_path.stat()
# TODO: consider checking if the file was just written to (which would mean that the file is still
# being copied). A segfault might happen in this edge case.
if prover.get_size() >= 30 and stat_info.st_size < 0.98 * expected_size:
log.warning(
f"Not farming plot {file_path}. Size is {stat_info.st_size / (1024**3)} GiB, but expected"
f" at least: {expected_size / (1024 ** 3)} GiB. We assume the file is being copied."
)
return new_provers
(
pool_public_key_or_puzzle_hash,
farmer_public_key,
local_master_sk,
) = parse_plot_info(prover.get_memo())
# Only use plots that correct keys associated with them
if self.farmer_public_keys is not None and farmer_public_key not in self.farmer_public_keys:
log.warning(f"Plot {file_path} has a farmer public key that is not in the farmer's pk list.")
self.no_key_filenames.add(file_path)
if not self.open_no_key_filenames:
return new_provers
if isinstance(pool_public_key_or_puzzle_hash, G1Element):
pool_public_key = pool_public_key_or_puzzle_hash
pool_contract_puzzle_hash = None
else:
assert isinstance(pool_public_key_or_puzzle_hash, bytes32)
pool_public_key = None
pool_contract_puzzle_hash = pool_public_key_or_puzzle_hash
if (
self.pool_public_keys is not None
and pool_public_key is not None
and pool_public_key not in self.pool_public_keys
):
log.warning(f"Plot {file_path} has a pool public key that is not in the farmer's pool pk list.")
self.no_key_filenames.add(file_path)
if not self.open_no_key_filenames:
return new_provers
stat_info = file_path.stat()
local_sk = master_sk_to_local_sk(local_master_sk)
plot_public_key: G1Element = ProofOfSpace.generate_plot_public_key(
local_sk.get_g1(), farmer_public_key, pool_contract_puzzle_hash is not None
)
with self.plot_filename_paths_lock:
if file_path.name not in self.plot_filename_paths:
self.plot_filename_paths[file_path.name] = (str(Path(prover.get_filename()).parent), set())
else:
self.plot_filename_paths[file_path.name][1].add(str(Path(prover.get_filename()).parent))
if len(self.plot_filename_paths[file_path.name][1]) > 0:
log.warning(
f"Have multiple copies of the plot {file_path} in "
f"{self.plot_filename_paths[file_path.name][1]}."
)
return new_provers
new_provers[file_path] = PlotInfo(
prover,
pool_public_key,
pool_contract_puzzle_hash,
plot_public_key,
stat_info.st_size,
stat_info.st_mtime,
)
with counter_lock:
result.loaded_plots += 1
result.loaded_size += stat_info.st_size
except Exception as e:
tb = traceback.format_exc()
log.error(f"Failed to open file {file_path}. {e} {tb}")
self.failed_to_open_filenames[file_path] = int(time.time())
return new_provers
log.info(f"Found plot {file_path} of size {new_provers[file_path].prover.get_size()}")
if self.show_memo:
plot_memo: bytes32
if pool_contract_puzzle_hash is None:
plot_memo = stream_plot_info_pk(pool_public_key, farmer_public_key, local_master_sk)
else:
plot_memo = stream_plot_info_ph(pool_contract_puzzle_hash, farmer_public_key, local_master_sk)
plot_memo_str: str = plot_memo.hex()
log.info(f"Memo: {plot_memo_str}")
return new_provers
return new_provers
def load_plots(
provers: Dict[Path, PlotInfo],
failed_to_open_filenames: Dict[Path, int],
farmer_public_keys: Optional[List[G1Element]],
pool_public_keys: Optional[List[G1Element]],
match_str: Optional[str],
root_path: Path,
open_no_key_filenames=False,
) -> Tuple[bool, Dict[Path, PlotInfo], Dict[Path, int], Set[Path]]:
start_time = time.time()
config_file = load_config(root_path, "config.yaml", "harvester")
changed = False
no_key_filenames: Set[Path] = set()
log.info(f'Searching directories {config_file["plot_directories"]}')
plot_filenames: Dict[Path, List[Path]] = get_plot_filenames(config_file)
all_filenames: List[Path] = []
for paths in plot_filenames.values():
all_filenames += paths
total_size = 0
new_provers: Dict[Path, PlotInfo] = {}
if match_str is not None:
log.info(
f'Only loading plots that contain "{match_str}" in the file or directory name'
)
for filename in all_filenames:
filename_str = str(filename)
if match_str is not None and match_str not in filename_str:
continue
if filename.exists():
if filename in failed_to_open_filenames and (
time.time() - failed_to_open_filenames[filename]) < 1200:
# Try once every 20 minutes to open the file
continue
if filename in provers:
stat_info = filename.stat()
if stat_info.st_mtime == provers[filename].time_modified:
total_size += stat_info.st_size
new_provers[filename] = provers[filename]
continue
try:
prover = DiskProver(str(filename))
expected_size = _expected_plot_size(
prover.get_size()) * UI_ACTUAL_SPACE_CONSTANT_FACTOR / 2.0
stat_info = filename.stat()
# TODO: consider checking if the file was just written to (which would mean that the file is still
# being copied). A segfault might happen in this edge case.
if prover.get_size(
) >= 30 and stat_info.st_size < 0.98 * expected_size:
log.warning(
f"Not farming plot {filename}. Size is {stat_info.st_size / (1024**3)} GiB, but expected"
f" at least: {expected_size / (1024 ** 3)} GiB. We assume the file is being copied."
)
continue
(
pool_public_key,
farmer_public_key,
local_master_sk,
) = parse_plot_info(prover.get_memo())
# Only use plots that correct keys associated with them
if farmer_public_keys is not None and farmer_public_key not in farmer_public_keys:
log.warning(
f"Plot {filename} has a farmer public key that is not in the farmer's pk list."
)
no_key_filenames.add(filename)
if not open_no_key_filenames:
continue
if pool_public_keys is not None and pool_public_key not in pool_public_keys:
log.warning(
f"Plot {filename} has a pool public key that is not in the farmer's pool pk list."
)
no_key_filenames.add(filename)
if not open_no_key_filenames:
continue
stat_info = filename.stat()
local_sk = master_sk_to_local_sk(local_master_sk)
plot_public_key: G1Element = ProofOfSpace.generate_plot_public_key(
local_sk.get_g1(), farmer_public_key)
new_provers[filename] = PlotInfo(
prover,
pool_public_key,
farmer_public_key,
plot_public_key,
local_sk,
stat_info.st_size,
stat_info.st_mtime,
)
total_size += stat_info.st_size
changed = True
except Exception as e:
tb = traceback.format_exc()
log.error(f"Failed to open file {filename}. {e} {tb}")
failed_to_open_filenames[filename] = int(time.time())
continue
log.info(
f"Found plot {filename} of size {new_provers[filename].prover.get_size()}"
)
log.info(
f"Loaded a total of {len(new_provers)} plots of size {total_size / (1024 ** 4)} TiB, in"
f" {time.time()-start_time} seconds")
return changed, new_provers, failed_to_open_filenames, no_key_filenames
def load_plots(
provers: Dict[Path, PlotInfo],
failed_to_open_filenames: Set[Path],
farmer_public_keys: Optional[List[G1Element]],
pool_public_keys: Optional[List[G1Element]],
root_path: Path,
open_no_key_filenames=False,
) -> Tuple[bool, Dict[Path, PlotInfo], Set[Path], Set[Path]]:
config_file = load_config(root_path, "config.yaml", "harvester")
changed = False
no_key_filenames: Set[Path] = set()
log.info(f'Searching directories {config_file["plot_directories"]}')
plot_filenames: Dict[Path, List[Path]] = get_plot_filenames(config_file)
all_filenames: List[Path] = []
for paths in plot_filenames.values():
all_filenames += paths
total_size = 0
for filename in all_filenames:
if filename in provers:
stat_info = filename.stat()
if stat_info.st_mtime == provers[filename].time_modified:
total_size += stat_info.st_size
continue
if filename in failed_to_open_filenames:
continue
if filename.exists():
try:
prover = DiskProver(str(filename))
(
pool_public_key,
farmer_public_key,
local_master_sk,
) = parse_plot_info(prover.get_memo())
# Only use plots that correct keys associated with them
if (farmer_public_keys is not None
and farmer_public_key not in farmer_public_keys):
log.warning(
f"Plot {filename} has a farmer public key that is not in the farmer's pk list."
)
no_key_filenames.add(filename)
if not open_no_key_filenames:
continue
if (pool_public_keys is not None
and pool_public_key not in pool_public_keys):
log.warning(
f"Plot {filename} has a pool public key that is not in the farmer's pool pk list."
)
no_key_filenames.add(filename)
if not open_no_key_filenames:
continue
stat_info = filename.stat()
local_sk = master_sk_to_local_sk(local_master_sk)
plot_public_key: G1Element = ProofOfSpace.generate_plot_public_key(
local_sk.get_g1(), farmer_public_key)
provers[filename] = PlotInfo(
prover,
pool_public_key,
farmer_public_key,
plot_public_key,
local_sk,
stat_info.st_size,
stat_info.st_mtime,
)
total_size += stat_info.st_size
changed = True
except Exception as e:
tb = traceback.format_exc()
log.error(f"Failed to open file {filename}. {e} {tb}")
failed_to_open_filenames.add(filename)
continue
log.info(
f"Found plot {filename} of size {provers[filename].prover.get_size()}"
)
log.info(
f"Loaded a total of {len(provers)} plots of size {total_size / (1024 ** 4)} TB"
)
return (changed, provers, failed_to_open_filenames, no_key_filenames)
