mirror of
https://github.com/exo-explore/exo.git
synced 2026-02-19 15:27:02 -05:00
Compare commits
4 Commits
meta-insta
...
leo/try-cl
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
31cd8b9d68 | ||
|
|
4c4c6ce99f | ||
|
|
42e1e7322b | ||
|
|
aa3f106fb9 |
@@ -20,6 +20,7 @@ from harness import (
|
||||
instance_id_from_instance,
|
||||
nodes_used_in_instance,
|
||||
resolve_model_short_id,
|
||||
run_planning_phase,
|
||||
settle_and_fetch_placements,
|
||||
wait_for_instance_gone,
|
||||
wait_for_instance_ready,
|
||||
@@ -962,6 +963,21 @@ Examples:
|
||||
|
||||
selected.sort(key=_placement_sort_key)
|
||||
preview = selected[0]
|
||||
|
||||
settle_deadline = (
|
||||
time.monotonic() + args.settle_timeout if args.settle_timeout > 0 else None
|
||||
)
|
||||
|
||||
print("Planning phase: checking downloads...", file=log)
|
||||
run_planning_phase(
|
||||
exo,
|
||||
full_model_id,
|
||||
preview,
|
||||
args.danger_delete_downloads,
|
||||
args.timeout,
|
||||
settle_deadline,
|
||||
)
|
||||
|
||||
instance = preview["instance"]
|
||||
instance_id = instance_id_from_instance(instance)
|
||||
sharding = str(preview["sharding"])
|
||||
|
||||
@@ -35,6 +35,7 @@ from harness import (
|
||||
instance_id_from_instance,
|
||||
nodes_used_in_instance,
|
||||
resolve_model_short_id,
|
||||
run_planning_phase,
|
||||
settle_and_fetch_placements,
|
||||
wait_for_instance_gone,
|
||||
wait_for_instance_ready,
|
||||
@@ -332,6 +333,20 @@ def main() -> int:
|
||||
if args.dry_run:
|
||||
return 0
|
||||
|
||||
settle_deadline = (
|
||||
time.monotonic() + args.settle_timeout if args.settle_timeout > 0 else None
|
||||
)
|
||||
|
||||
logger.info("Planning phase: checking downloads...")
|
||||
run_planning_phase(
|
||||
client,
|
||||
full_model_id,
|
||||
selected[0],
|
||||
args.danger_delete_downloads,
|
||||
args.timeout,
|
||||
settle_deadline,
|
||||
)
|
||||
|
||||
all_rows: list[dict[str, Any]] = []
|
||||
|
||||
for preview in selected:
|
||||
|
||||
150
bench/harness.py
150
bench/harness.py
@@ -282,6 +282,151 @@ def settle_and_fetch_placements(
|
||||
return selected
|
||||
|
||||
|
||||
def run_planning_phase(
|
||||
client: ExoClient,
|
||||
full_model_id: str,
|
||||
preview: dict[str, Any],
|
||||
danger_delete: bool,
|
||||
timeout: float,
|
||||
settle_deadline: float | None,
|
||||
) -> None:
|
||||
"""Check disk space and ensure model is downloaded before benchmarking."""
|
||||
# Get model size from /models
|
||||
models = client.request_json("GET", "/models") or {}
|
||||
model_bytes = 0
|
||||
for m in models.get("data", []):
|
||||
if m.get("hugging_face_id") == full_model_id:
|
||||
model_bytes = m.get("storage_size_megabytes", 0) * 1024 * 1024
|
||||
break
|
||||
|
||||
if not model_bytes:
|
||||
logger.warning(
|
||||
f"Could not determine size for {full_model_id}, skipping disk check"
|
||||
)
|
||||
return
|
||||
|
||||
# Get nodes from preview
|
||||
inner = unwrap_instance(preview["instance"])
|
||||
node_ids = list(inner["shardAssignments"]["nodeToRunner"].keys())
|
||||
runner_to_shard = inner["shardAssignments"]["runnerToShard"]
|
||||
|
||||
state = client.request_json("GET", "/state")
|
||||
downloads = state.get("downloads", {})
|
||||
node_disk = state.get("nodeDisk", {})
|
||||
|
||||
for node_id in node_ids:
|
||||
node_downloads = downloads.get(node_id, [])
|
||||
|
||||
# Check if model already downloaded on this node
|
||||
already_downloaded = any(
|
||||
"DownloadCompleted" in p
|
||||
and unwrap_instance(p["DownloadCompleted"]["shardMetadata"])["modelCard"][
|
||||
"modelId"
|
||||
]
|
||||
== full_model_id
|
||||
for p in node_downloads
|
||||
)
|
||||
if already_downloaded:
|
||||
continue
|
||||
|
||||
# Wait for disk info if settle_deadline is set
|
||||
disk_info = node_disk.get(node_id, {})
|
||||
backoff = _SETTLE_INITIAL_BACKOFF_S
|
||||
while not disk_info and settle_deadline and time.monotonic() < settle_deadline:
|
||||
remaining = settle_deadline - time.monotonic()
|
||||
logger.info(
|
||||
f"Waiting for disk info on {node_id} ({remaining:.0f}s remaining)..."
|
||||
)
|
||||
time.sleep(min(backoff, remaining))
|
||||
backoff = min(backoff * _SETTLE_BACKOFF_MULTIPLIER, _SETTLE_MAX_BACKOFF_S)
|
||||
state = client.request_json("GET", "/state")
|
||||
node_disk = state.get("nodeDisk", {})
|
||||
disk_info = node_disk.get(node_id, {})
|
||||
|
||||
if not disk_info:
|
||||
logger.warning(f"No disk info for {node_id}, skipping space check")
|
||||
continue
|
||||
|
||||
avail = disk_info.get("available", {}).get("inBytes", 0)
|
||||
if avail >= model_bytes:
|
||||
continue
|
||||
|
||||
if not danger_delete:
|
||||
raise RuntimeError(
|
||||
f"Insufficient disk on {node_id}: need {model_bytes // (1024**3)}GB, "
|
||||
f"have {avail // (1024**3)}GB. Use --danger-delete-downloads to free space."
|
||||
)
|
||||
|
||||
# Delete from smallest to largest
|
||||
completed = [
|
||||
(
|
||||
unwrap_instance(p["DownloadCompleted"]["shardMetadata"])["modelCard"][
|
||||
"modelId"
|
||||
],
|
||||
p["DownloadCompleted"]["totalBytes"]["inBytes"],
|
||||
)
|
||||
for p in node_downloads
|
||||
if "DownloadCompleted" in p
|
||||
]
|
||||
for del_model, size in sorted(completed, key=lambda x: x[1]):
|
||||
logger.info(f"Deleting {del_model} from {node_id} ({size // (1024**2)}MB)")
|
||||
client.request_json("DELETE", f"/download/{node_id}/{del_model}")
|
||||
avail += size
|
||||
if avail >= model_bytes:
|
||||
break
|
||||
|
||||
if avail < model_bytes:
|
||||
raise RuntimeError(f"Could not free enough space on {node_id}")
|
||||
|
||||
# Start downloads (idempotent)
|
||||
for node_id in node_ids:
|
||||
runner_id = inner["shardAssignments"]["nodeToRunner"][node_id]
|
||||
shard = runner_to_shard[runner_id]
|
||||
client.request_json(
|
||||
"POST",
|
||||
"/download/start",
|
||||
body={
|
||||
"targetNodeId": node_id,
|
||||
"shardMetadata": shard,
|
||||
},
|
||||
)
|
||||
logger.info(f"Started download on {node_id}")
|
||||
|
||||
# Wait for downloads
|
||||
start = time.time()
|
||||
while time.time() - start < timeout:
|
||||
state = client.request_json("GET", "/state")
|
||||
downloads = state.get("downloads", {})
|
||||
all_done = True
|
||||
for node_id in node_ids:
|
||||
done = any(
|
||||
"DownloadCompleted" in p
|
||||
and unwrap_instance(p["DownloadCompleted"]["shardMetadata"])[
|
||||
"modelCard"
|
||||
]["modelId"]
|
||||
== full_model_id
|
||||
for p in downloads.get(node_id, [])
|
||||
)
|
||||
failed = [
|
||||
p["DownloadFailed"]["errorMessage"]
|
||||
for p in downloads.get(node_id, [])
|
||||
if "DownloadFailed" in p
|
||||
and unwrap_instance(p["DownloadFailed"]["shardMetadata"])["modelCard"][
|
||||
"modelId"
|
||||
]
|
||||
== full_model_id
|
||||
]
|
||||
if failed:
|
||||
raise RuntimeError(f"Download failed on {node_id}: {failed[0]}")
|
||||
if not done:
|
||||
all_done = False
|
||||
if all_done:
|
||||
return
|
||||
time.sleep(1)
|
||||
|
||||
raise TimeoutError("Downloads did not complete in time")
|
||||
|
||||
|
||||
def add_common_instance_args(ap: argparse.ArgumentParser) -> None:
|
||||
ap.add_argument("--host", default=os.environ.get("EXO_HOST", "localhost"))
|
||||
ap.add_argument(
|
||||
@@ -325,3 +470,8 @@ def add_common_instance_args(ap: argparse.ArgumentParser) -> None:
|
||||
default=0,
|
||||
help="Max seconds to wait for the cluster to produce valid placements (0 = try once).",
|
||||
)
|
||||
ap.add_argument(
|
||||
"--danger-delete-downloads",
|
||||
action="store_true",
|
||||
help="Delete existing models from smallest to largest to make room for benchmark model.",
|
||||
)
|
||||
|
||||
@@ -19,7 +19,7 @@ class ConnectionUpdate:
|
||||
Whether this is a connection or disconnection event
|
||||
"""
|
||||
@property
|
||||
def peer_id(self) -> PeerId:
|
||||
def peer_id(self) -> builtins.str:
|
||||
r"""
|
||||
Identity of the peer that we have connected to or disconnected from.
|
||||
"""
|
||||
@@ -40,92 +40,22 @@ class Keypair:
|
||||
Identity keypair of a node.
|
||||
"""
|
||||
@staticmethod
|
||||
def generate_ed25519() -> Keypair:
|
||||
def generate() -> Keypair:
|
||||
r"""
|
||||
Generate a new Ed25519 keypair.
|
||||
"""
|
||||
@staticmethod
|
||||
def generate_ecdsa() -> Keypair:
|
||||
def from_bytes(bytes: bytes) -> Keypair:
|
||||
r"""
|
||||
Generate a new ECDSA keypair.
|
||||
"""
|
||||
@staticmethod
|
||||
def generate_secp256k1() -> Keypair:
|
||||
r"""
|
||||
Generate a new Secp256k1 keypair.
|
||||
"""
|
||||
@staticmethod
|
||||
def from_protobuf_encoding(bytes: bytes) -> Keypair:
|
||||
r"""
|
||||
Decode a private key from a protobuf structure and parse it as a `Keypair`.
|
||||
"""
|
||||
@staticmethod
|
||||
def rsa_from_pkcs8(bytes: bytes) -> Keypair:
|
||||
r"""
|
||||
Decode an keypair from a DER-encoded secret key in PKCS#8 `PrivateKeyInfo`
|
||||
format (i.e. unencrypted) as defined in [RFC5208].
|
||||
|
||||
[RFC5208]: https://tools.ietf.org/html/rfc5208#section-5
|
||||
"""
|
||||
@staticmethod
|
||||
def secp256k1_from_der(bytes: bytes) -> Keypair:
|
||||
r"""
|
||||
Decode a keypair from a DER-encoded Secp256k1 secret key in an `ECPrivateKey`
|
||||
structure as defined in [RFC5915].
|
||||
|
||||
[RFC5915]: https://tools.ietf.org/html/rfc5915
|
||||
"""
|
||||
@staticmethod
|
||||
def ed25519_from_bytes(bytes: bytes) -> Keypair: ...
|
||||
def to_protobuf_encoding(self) -> bytes:
|
||||
r"""
|
||||
Encode a private key as protobuf structure.
|
||||
"""
|
||||
def to_peer_id(self) -> PeerId:
|
||||
r"""
|
||||
Convert the `Keypair` into the corresponding `PeerId`.
|
||||
"""
|
||||
|
||||
@typing.final
|
||||
class Multiaddr:
|
||||
r"""
|
||||
Representation of a Multiaddr.
|
||||
"""
|
||||
@staticmethod
|
||||
def empty() -> Multiaddr:
|
||||
r"""
|
||||
Create a new, empty multiaddress.
|
||||
"""
|
||||
@staticmethod
|
||||
def with_capacity(n: builtins.int) -> Multiaddr:
|
||||
r"""
|
||||
Create a new, empty multiaddress with the given capacity.
|
||||
"""
|
||||
@staticmethod
|
||||
def from_bytes(bytes: bytes) -> Multiaddr:
|
||||
r"""
|
||||
Parse a `Multiaddr` value from its byte slice representation.
|
||||
"""
|
||||
@staticmethod
|
||||
def from_string(string: builtins.str) -> Multiaddr:
|
||||
r"""
|
||||
Parse a `Multiaddr` value from its string representation.
|
||||
"""
|
||||
def len(self) -> builtins.int:
|
||||
r"""
|
||||
Return the length in bytes of this multiaddress.
|
||||
"""
|
||||
def is_empty(self) -> builtins.bool:
|
||||
r"""
|
||||
Returns true if the length of this multiaddress is 0.
|
||||
Construct an Ed25519 keypair from secret key bytes
|
||||
"""
|
||||
def to_bytes(self) -> bytes:
|
||||
r"""
|
||||
Return a copy of this [`Multiaddr`]'s byte representation.
|
||||
Get the secret key bytes underlying the keypair
|
||||
"""
|
||||
def to_string(self) -> builtins.str:
|
||||
def to_node_id(self) -> builtins.str:
|
||||
r"""
|
||||
Convert a Multiaddr to a string.
|
||||
Convert the `Keypair` into the corresponding `PeerId` string, which we use as our `NodeId`.
|
||||
"""
|
||||
|
||||
@typing.final
|
||||
@@ -180,37 +110,6 @@ class NoPeersSubscribedToTopicError(builtins.Exception):
|
||||
def __repr__(self) -> builtins.str: ...
|
||||
def __str__(self) -> builtins.str: ...
|
||||
|
||||
@typing.final
|
||||
class PeerId:
|
||||
r"""
|
||||
Identifier of a peer of the network.
|
||||
|
||||
The data is a `CIDv0` compatible multihash of the protobuf encoded public key of the peer
|
||||
as specified in [specs/peer-ids](https://github.com/libp2p/specs/blob/master/peer-ids/peer-ids.md).
|
||||
"""
|
||||
@staticmethod
|
||||
def random() -> PeerId:
|
||||
r"""
|
||||
Generates a random peer ID from a cryptographically secure PRNG.
|
||||
|
||||
This is useful for randomly walking on a DHT, or for testing purposes.
|
||||
"""
|
||||
@staticmethod
|
||||
def from_bytes(bytes: bytes) -> PeerId:
|
||||
r"""
|
||||
Parses a `PeerId` from bytes.
|
||||
"""
|
||||
def to_bytes(self) -> bytes:
|
||||
r"""
|
||||
Returns a raw bytes representation of this `PeerId`.
|
||||
"""
|
||||
def to_base58(self) -> builtins.str:
|
||||
r"""
|
||||
Returns a base-58 encoded string of this `PeerId`.
|
||||
"""
|
||||
def __repr__(self) -> builtins.str: ...
|
||||
def __str__(self) -> builtins.str: ...
|
||||
|
||||
@typing.final
|
||||
class ConnectionUpdateType(enum.Enum):
|
||||
r"""
|
||||
|
||||
@@ -1,8 +1,6 @@
|
||||
use crate::ext::ResultExt as _;
|
||||
use libp2p::PeerId;
|
||||
use libp2p::identity::Keypair;
|
||||
use pyo3::prelude::{PyBytesMethods as _, PyModule, PyModuleMethods as _};
|
||||
use pyo3::types::PyBytes;
|
||||
use pyo3::types::{PyBytes, PyBytesMethods as _};
|
||||
use pyo3::{Bound, PyResult, Python, pyclass, pymethods};
|
||||
use pyo3_stub_gen::derive::{gen_stub_pyclass, gen_stub_pymethods};
|
||||
|
||||
@@ -18,142 +16,32 @@ pub struct PyKeypair(pub Keypair);
|
||||
impl PyKeypair {
|
||||
/// Generate a new Ed25519 keypair.
|
||||
#[staticmethod]
|
||||
fn generate_ed25519() -> Self {
|
||||
fn generate() -> Self {
|
||||
Self(Keypair::generate_ed25519())
|
||||
}
|
||||
|
||||
/// Generate a new ECDSA keypair.
|
||||
/// Construct an Ed25519 keypair from secret key bytes
|
||||
#[staticmethod]
|
||||
fn generate_ecdsa() -> Self {
|
||||
Self(Keypair::generate_ecdsa())
|
||||
}
|
||||
|
||||
/// Generate a new Secp256k1 keypair.
|
||||
#[staticmethod]
|
||||
fn generate_secp256k1() -> Self {
|
||||
Self(Keypair::generate_secp256k1())
|
||||
}
|
||||
|
||||
/// Decode a private key from a protobuf structure and parse it as a `Keypair`.
|
||||
#[staticmethod]
|
||||
fn from_protobuf_encoding(bytes: Bound<'_, PyBytes>) -> PyResult<Self> {
|
||||
let bytes = Vec::from(bytes.as_bytes());
|
||||
Ok(Self(Keypair::from_protobuf_encoding(&bytes).pyerr()?))
|
||||
}
|
||||
|
||||
/// Decode an keypair from a DER-encoded secret key in PKCS#8 `PrivateKeyInfo`
|
||||
/// format (i.e. unencrypted) as defined in [RFC5208].
|
||||
///
|
||||
/// [RFC5208]: https://tools.ietf.org/html/rfc5208#section-5
|
||||
#[staticmethod]
|
||||
fn rsa_from_pkcs8(bytes: Bound<'_, PyBytes>) -> PyResult<Self> {
|
||||
let mut bytes = Vec::from(bytes.as_bytes());
|
||||
Ok(Self(Keypair::rsa_from_pkcs8(&mut bytes).pyerr()?))
|
||||
}
|
||||
|
||||
/// Decode a keypair from a DER-encoded Secp256k1 secret key in an `ECPrivateKey`
|
||||
/// structure as defined in [RFC5915].
|
||||
///
|
||||
/// [RFC5915]: https://tools.ietf.org/html/rfc5915
|
||||
#[staticmethod]
|
||||
fn secp256k1_from_der(bytes: Bound<'_, PyBytes>) -> PyResult<Self> {
|
||||
let mut bytes = Vec::from(bytes.as_bytes());
|
||||
Ok(Self(Keypair::secp256k1_from_der(&mut bytes).pyerr()?))
|
||||
}
|
||||
|
||||
#[staticmethod]
|
||||
fn ed25519_from_bytes(bytes: Bound<'_, PyBytes>) -> PyResult<Self> {
|
||||
fn from_bytes(bytes: Bound<'_, PyBytes>) -> PyResult<Self> {
|
||||
let mut bytes = Vec::from(bytes.as_bytes());
|
||||
Ok(Self(Keypair::ed25519_from_bytes(&mut bytes).pyerr()?))
|
||||
}
|
||||
|
||||
/// Encode a private key as protobuf structure.
|
||||
fn to_protobuf_encoding<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyBytes>> {
|
||||
let bytes = self.0.to_protobuf_encoding().pyerr()?;
|
||||
/// Get the secret key bytes underlying the keypair
|
||||
fn to_bytes<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyBytes>> {
|
||||
let bytes = self
|
||||
.0
|
||||
.clone()
|
||||
.try_into_ed25519()
|
||||
.pyerr()?
|
||||
.secret()
|
||||
.as_ref()
|
||||
.to_vec();
|
||||
Ok(PyBytes::new(py, &bytes))
|
||||
}
|
||||
|
||||
/// Convert the `Keypair` into the corresponding `PeerId`.
|
||||
fn to_peer_id(&self) -> PyPeerId {
|
||||
PyPeerId(self.0.public().to_peer_id())
|
||||
}
|
||||
|
||||
// /// Hidden constructor for pickling support. TODO: figure out how to do pickling...
|
||||
// #[gen_stub(skip)]
|
||||
// #[new]
|
||||
// fn py_new(bytes: Bound<'_, PyBytes>) -> PyResult<Self> {
|
||||
// Self::from_protobuf_encoding(bytes)
|
||||
// }
|
||||
//
|
||||
// #[gen_stub(skip)]
|
||||
// fn __setstate__(&mut self, state: Bound<'_, PyBytes>) -> PyResult<()> {
|
||||
// *self = Self::from_protobuf_encoding(state)?;
|
||||
// Ok(())
|
||||
// }
|
||||
//
|
||||
// #[gen_stub(skip)]
|
||||
// fn __getstate__<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyBytes>> {
|
||||
// self.to_protobuf_encoding(py)
|
||||
// }
|
||||
//
|
||||
// #[gen_stub(skip)]
|
||||
// pub fn __getnewargs__<'py>(&self, py: Python<'py>) -> PyResult<(Bound<'py, PyBytes>,)> {
|
||||
// Ok((self.to_protobuf_encoding(py)?,))
|
||||
// }
|
||||
}
|
||||
|
||||
/// Identifier of a peer of the network.
|
||||
///
|
||||
/// The data is a `CIDv0` compatible multihash of the protobuf encoded public key of the peer
|
||||
/// as specified in [specs/peer-ids](https://github.com/libp2p/specs/blob/master/peer-ids/peer-ids.md).
|
||||
#[gen_stub_pyclass]
|
||||
#[pyclass(name = "PeerId", frozen)]
|
||||
#[derive(Debug, Clone)]
|
||||
#[repr(transparent)]
|
||||
pub struct PyPeerId(pub PeerId);
|
||||
|
||||
#[gen_stub_pymethods]
|
||||
#[pymethods]
|
||||
#[allow(clippy::needless_pass_by_value)]
|
||||
impl PyPeerId {
|
||||
/// Generates a random peer ID from a cryptographically secure PRNG.
|
||||
///
|
||||
/// This is useful for randomly walking on a DHT, or for testing purposes.
|
||||
#[staticmethod]
|
||||
fn random() -> Self {
|
||||
Self(PeerId::random())
|
||||
}
|
||||
|
||||
/// Parses a `PeerId` from bytes.
|
||||
#[staticmethod]
|
||||
fn from_bytes(bytes: Bound<'_, PyBytes>) -> PyResult<Self> {
|
||||
let bytes = Vec::from(bytes.as_bytes());
|
||||
Ok(Self(PeerId::from_bytes(&bytes).pyerr()?))
|
||||
}
|
||||
|
||||
/// Returns a raw bytes representation of this `PeerId`.
|
||||
fn to_bytes<'py>(&self, py: Python<'py>) -> Bound<'py, PyBytes> {
|
||||
let bytes = self.0.to_bytes();
|
||||
PyBytes::new(py, &bytes)
|
||||
}
|
||||
|
||||
/// Returns a base-58 encoded string of this `PeerId`.
|
||||
fn to_base58(&self) -> String {
|
||||
self.0.to_base58()
|
||||
}
|
||||
|
||||
fn __repr__(&self) -> String {
|
||||
format!("PeerId({})", self.to_base58())
|
||||
}
|
||||
|
||||
fn __str__(&self) -> String {
|
||||
self.to_base58()
|
||||
/// Convert the `Keypair` into the corresponding `PeerId` string, which we use as our `NodeId`.
|
||||
fn to_node_id(&self) -> String {
|
||||
self.0.public().to_peer_id().to_base58()
|
||||
}
|
||||
}
|
||||
|
||||
pub fn ident_submodule(m: &Bound<'_, PyModule>) -> PyResult<()> {
|
||||
m.add_class::<PyKeypair>()?;
|
||||
m.add_class::<PyPeerId>()?;
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
@@ -8,9 +8,10 @@ mod allow_threading;
|
||||
mod ident;
|
||||
mod networking;
|
||||
|
||||
use crate::ident::ident_submodule;
|
||||
use crate::ident::PyKeypair;
|
||||
use crate::networking::networking_submodule;
|
||||
use pyo3::prelude::PyModule;
|
||||
use pyo3::types::PyModuleMethods;
|
||||
use pyo3::{Bound, PyResult, pyclass, pymodule};
|
||||
use pyo3_stub_gen::define_stub_info_gatherer;
|
||||
|
||||
@@ -158,7 +159,7 @@ fn main_module(m: &Bound<'_, PyModule>) -> PyResult<()> {
|
||||
// TODO: for now this is all NOT a submodule, but figure out how to make the submodule system
|
||||
// work with maturin, where the types generate correctly, in the right folder, without
|
||||
// too many importing issues...
|
||||
ident_submodule(m)?;
|
||||
m.add_class::<PyKeypair>()?;
|
||||
networking_submodule(m)?;
|
||||
|
||||
// top-level constructs
|
||||
|
||||
@@ -8,7 +8,7 @@
|
||||
use crate::r#const::MPSC_CHANNEL_SIZE;
|
||||
use crate::ext::{ByteArrayExt as _, FutureExt, PyErrExt as _};
|
||||
use crate::ext::{ResultExt as _, TokioMpscReceiverExt as _, TokioMpscSenderExt as _};
|
||||
use crate::ident::{PyKeypair, PyPeerId};
|
||||
use crate::ident::PyKeypair;
|
||||
use crate::pyclass;
|
||||
use libp2p::futures::StreamExt as _;
|
||||
use libp2p::gossipsub;
|
||||
@@ -119,7 +119,7 @@ struct PyConnectionUpdate {
|
||||
|
||||
/// Identity of the peer that we have connected to or disconnected from.
|
||||
#[pyo3(get)]
|
||||
peer_id: PyPeerId,
|
||||
peer_id: String,
|
||||
|
||||
/// Remote connection's IPv4 address.
|
||||
#[pyo3(get)]
|
||||
@@ -251,7 +251,7 @@ async fn networking_task(
|
||||
// send connection event to channel (or exit if connection closed)
|
||||
if let Err(e) = connection_update_tx.send(PyConnectionUpdate {
|
||||
update_type: PyConnectionUpdateType::Connected,
|
||||
peer_id: PyPeerId(peer_id),
|
||||
peer_id: peer_id.to_base58(),
|
||||
remote_ipv4,
|
||||
remote_tcp_port,
|
||||
}).await {
|
||||
@@ -272,7 +272,7 @@ async fn networking_task(
|
||||
// send disconnection event to channel (or exit if connection closed)
|
||||
if let Err(e) = connection_update_tx.send(PyConnectionUpdate {
|
||||
update_type: PyConnectionUpdateType::Disconnected,
|
||||
peer_id: PyPeerId(peer_id),
|
||||
peer_id: peer_id.to_base58(),
|
||||
remote_ipv4,
|
||||
remote_tcp_port,
|
||||
}).await {
|
||||
|
||||
@@ -45,7 +45,7 @@ class Node:
|
||||
@classmethod
|
||||
async def create(cls, args: "Args") -> "Self":
|
||||
keypair = get_node_id_keypair()
|
||||
node_id = NodeId(keypair.to_peer_id().to_base58())
|
||||
node_id = NodeId(keypair.to_node_id())
|
||||
session_id = SessionId(master_node_id=node_id, election_clock=0)
|
||||
router = Router.create(keypair)
|
||||
await router.register_topic(topics.GLOBAL_EVENTS)
|
||||
|
||||
@@ -31,6 +31,7 @@ from exo.shared.types.openai_responses import (
|
||||
ResponseOutputText,
|
||||
ResponsesRequest,
|
||||
ResponsesResponse,
|
||||
ResponsesStreamEvent,
|
||||
ResponseTextDeltaEvent,
|
||||
ResponseTextDoneEvent,
|
||||
ResponseUsage,
|
||||
@@ -38,6 +39,11 @@ from exo.shared.types.openai_responses import (
|
||||
from exo.shared.types.text_generation import InputMessage, TextGenerationTaskParams
|
||||
|
||||
|
||||
def _format_sse(event: ResponsesStreamEvent) -> str:
|
||||
"""Format a streaming event as an SSE message."""
|
||||
return f"event: {event.type}\ndata: {event.model_dump_json()}\n\n"
|
||||
|
||||
|
||||
def _extract_content(content: str | list[ResponseContentPart]) -> str:
|
||||
"""Extract plain text from a content field that may be a string or list of parts."""
|
||||
if isinstance(content, str):
|
||||
@@ -219,13 +225,13 @@ async def generate_responses_stream(
|
||||
created_event = ResponseCreatedEvent(
|
||||
sequence_number=next(seq), response=initial_response
|
||||
)
|
||||
yield f"event: response.created\ndata: {created_event.model_dump_json()}\n\n"
|
||||
yield _format_sse(created_event)
|
||||
|
||||
# response.in_progress
|
||||
in_progress_event = ResponseInProgressEvent(
|
||||
sequence_number=next(seq), response=initial_response
|
||||
)
|
||||
yield f"event: response.in_progress\ndata: {in_progress_event.model_dump_json()}\n\n"
|
||||
yield _format_sse(in_progress_event)
|
||||
|
||||
# response.output_item.added
|
||||
initial_item = ResponseMessageItem(
|
||||
@@ -236,7 +242,7 @@ async def generate_responses_stream(
|
||||
item_added = ResponseOutputItemAddedEvent(
|
||||
sequence_number=next(seq), output_index=0, item=initial_item
|
||||
)
|
||||
yield f"event: response.output_item.added\ndata: {item_added.model_dump_json()}\n\n"
|
||||
yield _format_sse(item_added)
|
||||
|
||||
# response.content_part.added
|
||||
initial_part = ResponseOutputText(text="")
|
||||
@@ -247,7 +253,7 @@ async def generate_responses_stream(
|
||||
content_index=0,
|
||||
part=initial_part,
|
||||
)
|
||||
yield f"event: response.content_part.added\ndata: {part_added.model_dump_json()}\n\n"
|
||||
yield _format_sse(part_added)
|
||||
|
||||
accumulated_text = ""
|
||||
function_call_items: list[ResponseFunctionCallItem] = []
|
||||
@@ -281,7 +287,7 @@ async def generate_responses_stream(
|
||||
output_index=next_output_index,
|
||||
item=fc_item,
|
||||
)
|
||||
yield f"event: response.output_item.added\ndata: {fc_added.model_dump_json()}\n\n"
|
||||
yield _format_sse(fc_added)
|
||||
|
||||
# response.function_call_arguments.delta
|
||||
args_delta = ResponseFunctionCallArgumentsDeltaEvent(
|
||||
@@ -290,7 +296,7 @@ async def generate_responses_stream(
|
||||
output_index=next_output_index,
|
||||
delta=tool.arguments,
|
||||
)
|
||||
yield f"event: response.function_call_arguments.delta\ndata: {args_delta.model_dump_json()}\n\n"
|
||||
yield _format_sse(args_delta)
|
||||
|
||||
# response.function_call_arguments.done
|
||||
args_done = ResponseFunctionCallArgumentsDoneEvent(
|
||||
@@ -300,7 +306,7 @@ async def generate_responses_stream(
|
||||
name=tool.name,
|
||||
arguments=tool.arguments,
|
||||
)
|
||||
yield f"event: response.function_call_arguments.done\ndata: {args_done.model_dump_json()}\n\n"
|
||||
yield _format_sse(args_done)
|
||||
|
||||
# response.output_item.done
|
||||
fc_done_item = ResponseFunctionCallItem(
|
||||
@@ -315,7 +321,7 @@ async def generate_responses_stream(
|
||||
output_index=next_output_index,
|
||||
item=fc_done_item,
|
||||
)
|
||||
yield f"event: response.output_item.done\ndata: {fc_item_done.model_dump_json()}\n\n"
|
||||
yield _format_sse(fc_item_done)
|
||||
|
||||
function_call_items.append(fc_done_item)
|
||||
next_output_index += 1
|
||||
@@ -331,7 +337,7 @@ async def generate_responses_stream(
|
||||
content_index=0,
|
||||
delta=chunk.text,
|
||||
)
|
||||
yield f"event: response.output_text.delta\ndata: {delta_event.model_dump_json()}\n\n"
|
||||
yield _format_sse(delta_event)
|
||||
|
||||
# response.output_text.done
|
||||
text_done = ResponseTextDoneEvent(
|
||||
@@ -341,7 +347,7 @@ async def generate_responses_stream(
|
||||
content_index=0,
|
||||
text=accumulated_text,
|
||||
)
|
||||
yield f"event: response.output_text.done\ndata: {text_done.model_dump_json()}\n\n"
|
||||
yield _format_sse(text_done)
|
||||
|
||||
# response.content_part.done
|
||||
final_part = ResponseOutputText(text=accumulated_text)
|
||||
@@ -352,7 +358,7 @@ async def generate_responses_stream(
|
||||
content_index=0,
|
||||
part=final_part,
|
||||
)
|
||||
yield f"event: response.content_part.done\ndata: {part_done.model_dump_json()}\n\n"
|
||||
yield _format_sse(part_done)
|
||||
|
||||
# response.output_item.done
|
||||
final_message_item = ResponseMessageItem(
|
||||
@@ -363,7 +369,7 @@ async def generate_responses_stream(
|
||||
item_done = ResponseOutputItemDoneEvent(
|
||||
sequence_number=next(seq), output_index=0, item=final_message_item
|
||||
)
|
||||
yield f"event: response.output_item.done\ndata: {item_done.model_dump_json()}\n\n"
|
||||
yield _format_sse(item_done)
|
||||
|
||||
# Create usage from usage data if available
|
||||
usage = None
|
||||
@@ -388,4 +394,4 @@ async def generate_responses_stream(
|
||||
completed_event = ResponseCompletedEvent(
|
||||
sequence_number=next(seq), response=final_response
|
||||
)
|
||||
yield f"event: response.completed\ndata: {completed_event.model_dump_json()}\n\n"
|
||||
yield _format_sse(completed_event)
|
||||
|
||||
@@ -42,7 +42,7 @@ from exo.utils.channels import channel
|
||||
@pytest.mark.asyncio
|
||||
async def test_master():
|
||||
keypair = get_node_id_keypair()
|
||||
node_id = NodeId(keypair.to_peer_id().to_base58())
|
||||
node_id = NodeId(keypair.to_node_id())
|
||||
session_id = SessionId(master_node_id=node_id, election_clock=0)
|
||||
|
||||
ge_sender, global_event_receiver = channel[ForwarderEvent]()
|
||||
@@ -75,7 +75,7 @@ async def test_master():
|
||||
async with anyio.create_task_group() as tg:
|
||||
tg.start_soon(master.run)
|
||||
|
||||
sender_node_id = NodeId(f"{keypair.to_peer_id().to_base58()}_sender")
|
||||
sender_node_id = NodeId(f"{keypair.to_node_id()}_sender")
|
||||
# inject a NodeGatheredInfo event
|
||||
logger.info("inject a NodeGatheredInfo event")
|
||||
await local_event_sender.send(
|
||||
|
||||
@@ -30,7 +30,7 @@ class ConnectionMessage(CamelCaseModel):
|
||||
@classmethod
|
||||
def from_update(cls, update: ConnectionUpdate) -> "ConnectionMessage":
|
||||
return cls(
|
||||
node_id=NodeId(update.peer_id.to_base58()),
|
||||
node_id=NodeId(update.peer_id),
|
||||
connection_type=ConnectionMessageType.from_update_type(update.update_type),
|
||||
remote_ipv4=update.remote_ipv4,
|
||||
remote_tcp_port=update.remote_tcp_port,
|
||||
|
||||
@@ -221,7 +221,7 @@ def get_node_id_keypair(
|
||||
Obtain the :class:`PeerId` by from it.
|
||||
"""
|
||||
# TODO(evan): bring back node id persistence once we figure out how to deal with duplicates
|
||||
return Keypair.generate_ed25519()
|
||||
return Keypair.generate()
|
||||
|
||||
def lock_path(path: str | bytes | PathLike[str] | PathLike[bytes]) -> Path:
|
||||
return Path(str(path) + ".lock")
|
||||
@@ -235,12 +235,12 @@ def get_node_id_keypair(
|
||||
protobuf_encoded = f.read()
|
||||
|
||||
try: # if decoded successfully, save & return
|
||||
return Keypair.from_protobuf_encoding(protobuf_encoded)
|
||||
return Keypair.from_bytes(protobuf_encoded)
|
||||
except ValueError as e: # on runtime error, assume corrupt file
|
||||
logger.warning(f"Encountered error when trying to get keypair: {e}")
|
||||
|
||||
# if no valid credentials, create new ones and persist
|
||||
with open(path, "w+b") as f:
|
||||
keypair = Keypair.generate_ed25519()
|
||||
f.write(keypair.to_protobuf_encoding())
|
||||
f.write(keypair.to_bytes())
|
||||
return keypair
|
||||
|
||||
@@ -23,7 +23,7 @@ def _get_keypair_concurrent_subprocess_task(
|
||||
sem.release()
|
||||
# wait to be told to begin simultaneous read
|
||||
ev.wait()
|
||||
queue.put(get_node_id_keypair().to_protobuf_encoding())
|
||||
queue.put(get_node_id_keypair().to_bytes())
|
||||
|
||||
|
||||
def _get_keypair_concurrent(num_procs: int) -> bytes:
|
||||
|
||||
@@ -573,6 +573,13 @@ def main(
|
||||
case Shutdown():
|
||||
current_status = RunnerShuttingDown()
|
||||
logger.info("runner shutting down")
|
||||
|
||||
del inference_model, image_model, tokenizer, group
|
||||
mx.clear_cache()
|
||||
import gc
|
||||
|
||||
gc.collect()
|
||||
|
||||
event_sender.send(
|
||||
RunnerStatusUpdated(
|
||||
runner_id=runner_id, runner_status=current_status
|
||||
@@ -581,6 +588,7 @@ def main(
|
||||
event_sender.send(TaskAcknowledged(task_id=task.task_id))
|
||||
|
||||
current_status = RunnerShutdown()
|
||||
break
|
||||
case _:
|
||||
raise ValueError(
|
||||
f"Received {task.__class__.__name__} outside of state machine in {current_status=}"
|
||||
@@ -597,13 +605,6 @@ def main(
|
||||
event_sender.send(
|
||||
RunnerStatusUpdated(runner_id=runner_id, runner_status=current_status)
|
||||
)
|
||||
if isinstance(current_status, RunnerShutdown):
|
||||
del inference_model, image_model, tokenizer, group
|
||||
mx.clear_cache()
|
||||
import gc
|
||||
|
||||
gc.collect()
|
||||
break
|
||||
|
||||
|
||||
@cache
|
||||
|
||||
Reference in New Issue
Block a user