Examples: query, "exact match", wildcard*, wild?ard, wild*rd
Fuzzy search: cake~ (finds cakes, bake)
Term boost: "red velvet"^4, chocolate^2
Field grouping: tags:(+work -"fun-stuff")
Escaping: Escape characters +-&|!(){}[]^"~*?:\ with \, e.g. \+
Range search: properties.timestamp:[1587729413488 TO *] (inclusive), properties.title:{A TO Z}(excluding A and Z)
Combinations: chocolate AND vanilla, chocolate OR vanilla, (chocolate OR vanilla) NOT "vanilla pudding"
Field search: properties.title:"The Title" AND text
Answered
Hi All! I'Ve Been Using The Clearml Agent In Tandem With My Self-Hosted Clearml Server On My Own Hardware. I'M Noticing That The Files Created By The Agent, E.G. Model Checkpoints, Debug Samples, Etc, Are Read Only To To The Agent'S Group. I'M Wondering I

Hi all! I've been using the ClearML agent in tandem with my self-hosted ClearML server on my own hardware. I'm noticing that the files created by the agent, e.g. model checkpoints, debug samples, etc, are read only to to the Agent's group. I'm wondering if there's a way to change the permissions when it writes files so that all users in the same group have rwx access. Reason is, I have other users who access my hardware, but they are not admin level and I restrict sudo access from tem. This inhibits them from doing things like deleting files which were made by the Agent, and the only solution so far is for me to sudo rm them myself.

  
  
Posted 4 months ago
Votes Newest

Answers 6


You can either set your user permission to allow group write by default ?
Or maybe create a dedicated user with group write permission and run the agent with that user ?

  
  
Posted 4 months ago

Hi @<1600661428556009472:profile|HighCoyote66> , are you running the agent in docker mode?

  
  
Posted 4 months ago

This is probably why all generated files are under the same user's privileges, since the agent simply spawns the task process using the same user it was executed with - that's one reason why running in a docker container helps 🙂

  
  
Posted 4 months ago

No, I'm using conda environment for execution. I've set the package manager to conda, and provided the path to a pre-built conda environment. Hence, I've also set conda_env_as_base_docker to true as well.

  
  
Posted 4 months ago

Makes sense. Howeveer, I'm not able to get docker on my distro, rather I am usng apptainer. Do you know of any alternatives?

  
  
Posted 4 months ago

I see, thank you guys!

  
  
Posted 4 months ago
484 Views
6 Answers
4 months ago
4 months ago
Tags