fix(python-sdk): stop leaking per-call proxy pools in volume content clients

[mishushakov]

The Python volume content client factories passed both proxy and the shared cached transport to httpx, so with a proxy configured (e.g. Volume.connect(volume_id, proxy="http://user:pass@127.0.0.1:8080")), every volume operation mounted a fresh, never-closed proxy transport that bypassed the cached connection pool. The client-level proxy argument is now dropped — the proxy is already baked into the cached transport, so proxied requests keep working but reuse one pooled transport per thread/event loop.

The volume transports also gained connect-level retries (E2B_CONNECTION_RETRIES, default 3), matching the core API and envd transports. Includes a changeset for a @e2b/python-sdk patch release.

🤖 Generated with Claude Code

[cursor]

PR Summary

Low Risk
Targeted httpx client wiring and transport retry parity; proxied volume traffic should behave the same but reuse pooled connections.

Overview
Fixes a connection-pool leak when volume content clients are used with a proxy: httpx_args no longer passes proxy alongside the shared cached transport, since the proxy is already configured on that transport. Passing both caused httpx to mount a new, never-closed proxy transport on every client creation.

Sync and async volume transports now use connection_retries (E2B_CONNECTION_RETRIES, default 3) at connect time, aligned with the core API and envd transports. Includes a @e2b/python-sdk patch changeset.

^{Reviewed by Cursor Bugbot for commit f5a3bb4. Bugbot is set up for automated code reviews on this repo. Configure here.}

[changeset-bot]

🦋 Changeset detected

Latest commit: f5a3bb4

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@e2b/python-sdk	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[github-actions]

Package Artifacts

Built from 361e3e7. Download artifacts from this workflow run.

JS SDK (e2b@2.32.1-volumes-unclosed-connections.0):

npm install ./e2b-2.32.1-volumes-unclosed-connections.0.tgz

CLI (@e2b/cli@2.13.2-volumes-unclosed-connections.0):

npm install ./e2b-cli-2.13.2-volumes-unclosed-connections.0.tgz

Python SDK (e2b==2.31.0+volumes.unclosed.connections):

pip install ./e2b-2.31.0+volumes.unclosed.connections-py3-none-any.whl

[claude]

LGTM — small, symmetric fix matching the pattern already used by the core API/envd transports (proxy dropped when transport is passed; connect retries added).

Extended reasoning...

Overview

This PR fixes a connection-pool leak in the Python SDK's volume content clients. Only 3 files change: a changeset and the sync/async volume client factories. The substantive change is ~5 lines in each of the two client files: drop "proxy": config.proxy from httpx_args (the proxy is already baked into the cached transport, so passing both caused httpx to mount a fresh, never-closed proxy transport on every client creation), and add retries=connection_retries to the transport constructor.

Security risks

None. The proxy is still honored via the cached transport — proxied traffic keeps working. No auth, crypto, or permission surface is touched.

Level of scrutiny

Low. This is a targeted bug fix that mirrors an established pattern in the same codebase: e2b/api/__init__.py:209-216 already sets httpx_args["proxy"] only when no transport is provided, and tests/test_api_client_transport.py:43,231 already asserts "proxy" not in api_client._httpx_args when a transport is passed. The connection_retries constant (E2B_CONNECTION_RETRIES, default 3) is defined in e2b/api/__init__.py:71 and already used by both the API and envd transports.

Other factors

Sync and async implementations are updated symmetrically as required by CLAUDE.md. A patch-level changeset is included. Cursor Bugbot flagged this as low risk. No outstanding reviewer comments.