Skip to content

fix(v1): detect env-server worker death instead of hanging #2131

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
eligotts wants to merge 7 commits into
base: main
Choose a base branch
from
+380 −92
Open

fix(v1): detect env-server worker death instead of hanging #2131

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
8924338
fix(v1): detect env-server worker death instead of hanging
Jul 24, 2026
25bd48d
fix(v1): keep the broker alive on spawn failure, drain replies before…
Jul 24, 2026
f0c87f2
fix(v1): retry the replacement spawn when the pool is left empty
Jul 24, 2026
bb764b6
Merge main and address env-server review feedback
eligotts Jul 28, 2026
6b4ad69
fix(v1): gate replacement workers on readiness
eligotts Jul 28, 2026
ca9fbbf
fix(v1): bound failed pool dispatch
eligotts Jul 28, 2026
40154d6
fix(v1): isolate worker recovery budgets
eligotts Jul 28, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 7 additions & 3 deletions verifiers/v1/cli/eval/runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -109,7 +109,12 @@ async def run_eval_server(config: EvalConfig) -> list[Episode]:

from verifiers.v1.utils.logging import setup_logging
from verifiers.v1.configs.cli.env import pool_serve_kwargs
from verifiers.v1.serve import EnvClient, env_config_data, serve_env
from verifiers.v1.serve import (
EnvClient,
env_config_data,
serve_env,
wait_for_address,
)

legacy = config.is_legacy
server_kwargs = (
Expand Down Expand Up @@ -155,9 +160,8 @@ async def run_eval_server(config: EvalConfig) -> list[Episode]:
proc.start()
child_conn.close() # the child holds its end; we keep parent_conn so our exit closes it
try:
address = await asyncio.to_thread(address_queue.get, timeout=600)
address = await wait_for_address(proc, address_queue, log_file=log_file)
Comment thread
cursor[bot] marked this conversation as resolved.
client = EnvClient(address=address)
await client.wait_for_server_startup(timeout=600)
# A v1 run dispatches — and resumes — tasks by content: the client owns them,
# and `resume.task_key` is their identity. Only the legacy bridge is addressed
# by dataset row (its dataset lives server-side, reported via `info`), and
Expand Down
10 changes: 9 additions & 1 deletion verifiers/v1/serve/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,11 @@
from verifiers.v1.serve.client import EnvClient
from verifiers.v1.serve.pool import EnvServerPool, env_config_data, serve_env
from verifiers.v1.serve.pool import (
ENV_SERVER_SPAWN_TIMEOUT,
EnvServerPool,
env_config_data,
serve_env,
wait_for_address,
)
from verifiers.v1.serve.server import EnvServer
from verifiers.v1.serve.types import (
HealthRequest,
Expand All @@ -17,6 +23,8 @@
"EnvServerPool",
"serve_env",
"env_config_data",
"wait_for_address",
"ENV_SERVER_SPAWN_TIMEOUT",
"EnvClient",
"HealthRequest",
"HealthResponse",
Expand Down
Loading
Loading