Sandbox lifecycle: templates, timeouts and kill

A sandbox starts from a template, runs until its timeout or until you kill it, and can have its timeout extended. The defaults are in the SDK.

from e2b_code_interpreter import Sandbox
from e2b_code_interpreter.constants import DEFAULT_TIMEOUT

print("template:", Sandbox.default_template)
print("sandbox timeout:", Sandbox.default_sandbox_timeout, "seconds")
print("run_code timeout:", DEFAULT_TIMEOUT, "seconds")

The code interpreter's default template is code-interpreter-v1: Linux with a Jupyter kernel and common data libraries. A sandbox lives 300 seconds unless you say otherwise, and a single run_code may run for 300 seconds.

sandbox = Sandbox.create(timeout=60)      # alive for 60 seconds
sandbox.set_timeout(120)                  # now 120 seconds from this call
print(sandbox.sandbox_id)
print(sandbox.get_info())
sandbox.kill()

• timeout is in seconds in Python, and milliseconds, timeoutMs, in the JavaScript SDK.
• set_timeout restarts the countdown; an app calls it while a user is still active.
• sandbox_id identifies it; Sandbox.connect(sandbox_id) reattaches from another process.
• kill stops it immediately. An expired sandbox is stopped for you, or paused if you configured lifecycle options.

import inspect

from e2b_code_interpreter import Sandbox

for name in ["timeout", "allow_internet_access", "envs", "metadata", "template"]:
    print(name, "=", inspect.signature(Sandbox.create).parameters[name].default)

These are create's defaults, read from the installed SDK: no custom timeout, internet access on, no environment variables or metadata, the default template. Lesson 7 changes the internet default.