@<1523701994743664640:profile|AppetizingMouse58>

@<1523701994743664640:profile|AppetizingMouse58> Hi! Thank you for a quick response! Should I edit the configuration file directly in apiserver container like in screenshot? Is there a better, more consistent way to do that (I don't want to edit this config again if I recreate container for some reason) ?

@<1523701994743664640:profile|AppetizingMouse58> @<1523701087100473344:profile|SuccessfulKoala55> It worked as intended! Thank you once again :hugging_face:

@<1523701087100473344:profile|SuccessfulKoala55> Thank you! I think I've done something wrong (It doesn't delete now). Correct me please

• I've created file on clearml host machine /opt/clearml/config/services.conf with content as in the pic
• Restarted container with docker restart async-delete
  

Hi! I've faced the same problem and solved it like in this thread

None

In short - you have to map config folder to async_delete container like in apiserver container.

@<1523703097560403968:profile|CumbersomeCormorant74> Thank you! I'll try this and let you know if that helped

It worked! Thank you once again!

Thank you @<1523701070390366208:profile|CostlyOstrich36> Do you mean that I should edit clearml.conf and add something like this in agent section?

ok, thanks a lot! i need that package for all jobs

Thanks, I figured it out that it's enough to mount configmap into clearml pods

Here's my custom-values.yaml

# -- Private image registry configuration
imageCredentials:
  # -- Use private authentication mode
  enabled: true
  # -- If this is set, chart will not generate a secret but will use what is defined here
  existingSecret: ""
  # -- Registry name
  registry: gitlab.my-company
  # -- Registry username
  username: gitlab+deploy-token
  # -- Registry password
  password: token
...

@<1570583237065969664:profile|AdorableCrocodile14> Hi, probably you have dataset or pipeline in this project

@<1523701994743664640:profile|AppetizingMouse58> Thanks a lot! That worked ☺

@<1523701087100473344:profile|SuccessfulKoala55> Mate, any news on my issue from the team?

If you re facing problem with getting elastic search image, then I'd recommend using vpn to download it and then push it to your own docker-hub repo (in case you don't have vpn on server machine)

/mnt/data/k8s-pvs/clearml-agent-pv is a mounted folder

pip-download-cache folder is created but it stays empty

my current agent config for worker looks like the following now

agent {
      worker_id: "k8s-service-worker"
      worker_name: "k8s-service-worker"
      git_user: "gitlab+deploy-token-4"
      git_pass: "xxx"
      docker_force_pull: true
      package_manager.system_site_packages: false
      venvs_dir: ~/.cache/venvs-builds
      venvs_cache: {
          max_entries: 5
          free_space_threshold_gb: 10.0
          path: /mnt/data/k8s-pvs/clearml-agent-pv/venvs-cache
      },
 ...

@<1523701070390366208:profile|CostlyOstrich36> Could you please suggest something?

Hi! Try excluding “http/https” from the uri in the config. Have you configured your client correctly?Try adding “output_uri” (or smth like that) when initializing your Task.

Also worth to mention that I'm using helm chart 7.11.4

@<1523701070390366208:profile|CostlyOstrich36> Hello, mate!
Here's my values.yaml cleaned from keys.

Alright I've solved the issue by myself. The problem was that my agent used internal k8s domain names. I've changed them (in agent's helm values) to real ones, added certificates, set REQUESTS_CA_BUNDLE env var to point to my custom ca cert bundle and voila - debug samples are shown

Here's what I see. And the url of the image is k8s internal url
like None

If I substitute the internal address part ( None with actual address I can see the image ( None )
![image](https://clearml-web-assets.s3.amazonaws.com/scoold/images/TT9ATQXJ5-F0...