pytorch/test/distributed/checkpoint/test_fsspec.py

# Owner(s): ["oncall: distributed"]

import shutil
import tempfile
from collections.abc import Callable
from functools import wraps
from typing import Any, Optional

import torch
import torch.distributed as dist
import torch.distributed.checkpoint as dcp
import torch.nn as nn
from torch.distributed.checkpoint._fsspec_filesystem import (
    FileSystem,
    FsspecReader,
    FsspecWriter,
)
from torch.distributed.checkpoint.optimizer import load_sharded_optimizer_state_dict
from torch.distributed.checkpoint.utils import CheckpointException
from torch.distributed.fsdp import FullyShardedDataParallel as FSDP
from torch.distributed.fsdp.fully_sharded_data_parallel import StateDictType
from torch.testing._internal.common_distributed import (
    requires_accelerator_dist_backend,
    skip_if_lt_x_gpu,
)
from torch.testing._internal.common_utils import run_tests, TestCase
from torch.testing._internal.distributed._shard.sharded_tensor import (
    ShardedTensorTestBase,
    with_comms,
)


device_type = acc.type if (acc := torch.accelerator.current_accelerator()) else "cpu"
BACKEND = torch.distributed.get_default_backend_for_device(device_type)


def with_temp_dir(
    func: Optional[Callable] = None,
) -> Optional[Callable]:
    """
    Wrapper to initialize temp directory for distributed checkpoint.
    """
    assert func is not None

    @wraps(func)
    def wrapper(self, *args: tuple[object], **kwargs: dict[str, Any]) -> None:
        # Only create temp_dir when rank is 0 (or no pg)
        if not dist.is_initialized() or dist.get_rank() == 0:
            temp_dir = tempfile.mkdtemp()
            print(f"Using temp directory: {temp_dir}")
        else:
            temp_dir = ""
        object_list = [temp_dir]

        # Broadcast temp_dir to all the other ranks
        if dist.is_initialized():
            dist.broadcast_object_list(object_list)
        self.temp_dir = object_list[0]

        try:
            func(self, *args, **kwargs)
        finally:
            if not dist.is_initialized() or dist.get_rank() == 0:
                shutil.rmtree(self.temp_dir, ignore_errors=True)

    return wrapper


class MyTestModule(torch.nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.net1 = nn.Sequential(nn.Linear(8, 16), nn.ReLU())
        self.net2 = nn.Sequential(nn.Linear(16, 32), nn.ReLU())
        self.net3 = nn.Linear(32, 64)
        self.net4 = nn.Sequential(nn.ReLU(), nn.Linear(64, 8))

    def forward(self, x):
        return self.net4(self.net3(self.net2(self.net1(x))))


class TestFSSpec(ShardedTensorTestBase):
    @property
    def world_size(self) -> int:
        return 2

    @with_comms(backend=BACKEND, init_rpc=False)
    @requires_accelerator_dist_backend()
    @skip_if_lt_x_gpu(2)
    @with_temp_dir
    def test_fsspec(self):
        CHECKPOINT_DIR = self.temp_dir

        model = FSDP(MyTestModule().to(device_type))
        optim = torch.optim.Adam(model.parameters(), lr=0.1)
        model(torch.rand(8, 8, device=dist.get_rank())).sum().backward()
        optim.step()

        with FSDP.state_dict_type(model, StateDictType.SHARDED_STATE_DICT):
            state_dict = {
                "model": model.state_dict(),
                "optim": FSDP.optim_state_dict(model, optim),
            }

            dcp.save(
                state_dict=state_dict,
                storage_writer=FsspecWriter(CHECKPOINT_DIR),
                planner=dcp.DefaultSavePlanner(),
            )

        model_2 = FSDP(MyTestModule().to(device_type))
        optim_2 = torch.optim.Adam(model_2.parameters(), lr=0.1)

        with FSDP.summon_full_params(model):
            with FSDP.summon_full_params(model_2):
                for n_p1, n_p2 in zip(
                    model.named_parameters(), model_2.named_parameters()
                ):
                    self.assertNotEqual(n_p1[1], n_p2[1])

        # now load the model and ensure the values are the same
        with FSDP.state_dict_type(model_2, StateDictType.SHARDED_STATE_DICT):
            state_dict = {
                "model": model_2.state_dict(),
            }

            dcp.load(
                state_dict=state_dict,
                storage_reader=FsspecReader(CHECKPOINT_DIR),
                planner=dcp.DefaultLoadPlanner(),
            )
            model_2.load_state_dict(state_dict["model"])

            optim_state = load_sharded_optimizer_state_dict(
                model_state_dict=state_dict["model"],
                optimizer_key="optim",
                storage_reader=FsspecReader(CHECKPOINT_DIR),
            )

            flattened_osd = FSDP.optim_state_dict_to_load(
                model_2, optim_2, optim_state["optim"]
            )
            optim_2.load_state_dict(flattened_osd)

        with FSDP.summon_full_params(model):
            with FSDP.summon_full_params(model_2):
                for n_p1, n_p2 in zip(
                    model.named_parameters(), model_2.named_parameters()
                ):
                    self.assertEqual(n_p1[1], n_p2[1])

        def opt_at(opt, idx):
            return list(iter(opt.state.values()))[idx]

        # Adam lazily creates its state
        self.assertEqual(opt_at(optim, 0)["exp_avg"], opt_at(optim_2, 0)["exp_avg"])
        self.assertEqual(
            opt_at(optim, 0)["exp_avg_sq"], opt_at(optim_2, 0)["exp_avg_sq"]
        )

    @with_comms(backend=BACKEND, init_rpc=False)
    @requires_accelerator_dist_backend()
    @skip_if_lt_x_gpu(2)
    @with_temp_dir
    def test_overwrite(self):
        t1, t2 = torch.randn(10), torch.randn(10)

        dcp.save(
            {"random": t1}, storage_writer=FsspecWriter(self.temp_dir, overwrite=False)
        )
        dcp.save(
            {"random": t2}, storage_writer=FsspecWriter(self.temp_dir, overwrite=True)
        )

        sd = {"random": torch.zeros(10)}
        dcp.load(sd, checkpoint_id=self.temp_dir)
        self.assertTrue(torch.allclose(sd["random"], t2))

        with self.assertRaisesRegex(
            CheckpointException, ".*Checkpoint already exists.*"
        ):
            dcp.save(
                {"random": t2},
                storage_writer=FsspecWriter(self.temp_dir, overwrite=False),
            )


class TestFileSystem(TestCase):
    @with_temp_dir
    def test_remove_on_fail(self):
        fs = FileSystem()
        path = fs.init_path(self.temp_dir)

        write_file = fs.concat_path(path, "writeable")
        with self.assertRaises(OSError):
            with fs.create_stream(write_file, "w") as s:
                s.write("aaa")
                raise OSError("fail")
        self.assertFalse(fs.exists(write_file))

        read_file = fs.concat_path(path, "readable")
        with fs.create_stream(read_file, "w") as s:
            s.write("bbb")
        self.assertTrue(fs.exists(read_file))

        with self.assertRaises(OSError):
            with fs.create_stream(read_file, "r") as s:
                raise OSError("fail")
        self.assertTrue(fs.exists(read_file))


if __name__ == "__main__":
    run_tests()