[HTCondor-users] condor_submit pull policy for container universe vs. docker universe

[Matthew Feickert <matthew.feickert@xxxxxxxx>]

Hi All,

In https://github.com/UW-Madison-DSI/pixi-docker-chtc/pull/17 I am converting (among other things) from using the condor_submit "Docker universe" set of options to the "container universe" set of options. In doing so I noticed that my job raised

```

WARNING: the line 'container_pull_policy = missing' was unused by condor_submit. Is it a typo

```

Checking the condor_submit options docs, for the Docker Universe set of commands

https://htcondor.readthedocs.io/en/latest/man-pages/condor_submit.html#docker_pull_policy

> `docker_pull_policy = < always >`:

> if `docker_pull_policy` is set to _always_, when a docker universe job starts on a execution point, the option "–pull always" will be passed to the docker run command. This only impacts EPs which already have a locally cached version of the image. With this option, docker will always check with the repo to see if the cached version is out of date. This requires more network connectivity, and may cause docker hub to throttle future pull requests. It is generally recommended to never mutate docker image tag name, and avoid needing this option.

but then for the container universe set of commands there is no `container_pull_policy`. There is

https://htcondor.readthedocs.io/en/latest/man-pages/condor_submit.html#transfer_container

> `transfer_container = < True | False >`

> A boolean value that defaults to True. When `false`, sif container images and expanded directories are assumed to be pre-staged on the EP, and HTCondor will not attempt to transfer them.

but that sounds like it might be different(?).

Do you know if this means that there is no way to overwrite the container cache if the container universe is used?

Best,

Matthew

--
Matthew Feickert
University of Wisconsin-Madison

Book time to meet with me