OSERROR 12 when calling readline.write_history

[PhillipDowney]

Occasional error experienced by customer

/python3.11/site-packages/openad/app/main.py", line 526, in postloop
readline.write_history_file(self.histfile)
OSError: [Errno 12] Cannot allocate memory

Appears to be environmental on linux

[PhillipDowney]

Workaround being tested in

pip install git+https://github.com/acceleratedscience/openad-toolkit@bug_112 --no-cache

[PhillipDowney]

Waiting for user feedback

[darnells]

I have encountered the readline.write_history crash after applying the above patch. I am running in a vscode devcontainer project on Windows 11 Enterprise (23H2) with an up-to-date WSL platform. Podman Engine 5.4.2 is running within the stock FedoraLinux-42 image released on May 6, 2025. Here are the configuration files needed to reproduce my environment.

devcontainer.json

Dockerfile.txt

requirements.txt

[darnells]

This exception was thrown after processing between 3,185 and 3,430 antigen/light/heavy chain groupings using the async batch command and polling every 120 seconds after submission.

{service} generate with {model} data for {antigen_chain_list} USING ( hchain_seq={heavy_chain_list} lchain_seq={light_chain_list} ) async

Traceback (most recent call last):
  File "/workspaces/openad/example.py", line 213, in <module>
    main(parser.parse_args())
  File "/workspaces/openad/example.py", line 202, in main
    run_inference_batch(hndl, seqdata, cache)
  File "/workspaces/openad/example.py", line 105, in run_inference_batch
    retval = hndl.request(
             ^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/openad/api.py", line 53, in request
    result = self.main_app.api_remote(command, self.context_cache, api_variable)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/openad/app/main.py", line 1103, in api_remote
    magic_prompt.postloop()
  File "/usr/local/lib/python3.11/site-packages/openad/app/main.py", line 526, in postloop
    readline.write_history_file(self.histfile)
OSError: [Errno 12] Cannot allocate memory

[PhillipDowney]

thanks Steve, We have implemented a fix we are testing in Main pull request #138. We have gone to truncating history to 50, to avoid all possible memory scenarios. Will Update when regression testing finished.

[PhillipDowney]

@darnells We have been running tests in batch with combinations of 3600 + combinations per request. Using a derivations of your Rocky Dockerfile, poling every .2 of a second for a response , this seems to be working without replication of the error on a single handle.

Hoping this week to release a new Main version of the toolkit.

[PhillipDowney]

Problem addressed, final commit #143