#Idle artifact daemon consumes sustained CPU/RAM without active artifact task

1 messages · Page 1 of 1 (latest)

ornate shaleBOT
#

Reported by @alpine field

Bug Report: Idle artifact daemon consumes sustained CPU/RAM without active artifact task
`Steps to Reproduce`
  1. Start a normal validation/session workflow without requesting PDF, DOCX, PPTX, slide, or rich artifact generation.
  2. Inspect running processes.
  3. Observe artifact_tool_rpc_daemon-bun under presentation_artifact_tool.
  4. Sample CPU/RAM before and after 1+ minute.
  5. CPU remains around 27% and RAM around 463 MB while no artifact task is active.
`Expected Result`

Artifact tooling should lazy-start only when rich artifact work is requested, remain near 0% CPU while idle, expose task ownership, and shut down after a short idle timeout. Normal validation, filesystem, code, text, JSON, CSV, log, zip, or tar.gz workflows should not spawn or retain a high-CPU artifact daemon.

`Actual Result`

artifact_tool_rpc_daemon-bun remains active with sustained CPU usage around 27% and RSS around 463 MB for 1+ minute without an active artifact task. This consumes CPU/RAM, may prevent idle power states, adds cooling/load overhead, contaminates validation timing, and can cause unrelated workflows to slow down or time out.

`Environment`

ChatGPT sandbox; Python 3.13 pyvenv; presentation_artifact_tool daemon

#
Additional Information

Please provide relevant details to help resolve the issue, such as:

  • ChatGPT Shared Link (if applicable).
  • Screenshots or videos demonstrating the problem.

-# ➜ Need to contact support? Visit the OpenAI Help Center.

alpine field
#

it's costing open ai roughly $35k a day per gigawatt per center in wasted electricity at current rates.