logger.report_media( title=name_title, series="Nan", iteration=0, local_path=fig_nan, delete_after_upload=delete_after_upload, ) clearml_task.upload_artifact( name=name_title, artifact_object=fig_nan, wait_on_upload=True, )

tagging @<1523701205467926528:profile|AgitatedDove14> here just in case 😅

they are taking longer than 30 secs, but admittedly not much longer: 1-3 minutes

Not sure why it tries to establish some http connection, or why it's / ...

okay. what do you suggest?

Ok, going to ask the server admins, will keep you posted, thanks!

(base) emilio@unicorn:~$ docker version Client: Docker Engine - Community Version: 19.03.13 API version: 1.40 Go version: go1.13.15 Git commit: 4484c46d9d Built: Wed Sep 16 17:02:36 2020 OS/Arch: linux/amd64 Experimental: false (base) emilio@unicorn:~$ docker-compose --version docker-compose version 1.17.1, build unknown

I have done this but I remember someone once told me this could be an issue... Or I could be misremembering. I just wanted to double check

right, seems to have worked now!

api_server:

when an agent launches a task, it builds a venv, copies the code, runs it, etc. in my case, the code writes files (such as data it downloaded, or model files, etc) and writes them in subfolders. I'm interested in recovering the entire folder structure.

this is because if I run a different task, everything from the previous task is overwritten.

furthermore, I need the folder structure for other things downstream

and this should force the model to be uploaded?

Hi SuccessfulKoala55 I am having some issues with this. I have put a concurrency limit of 2 and I can see 3 workers running

see this for example

If 3e5962dd is the commit it's trying to clone, it doesn't exist because I deleted it.
it should be cloning a more up-to-date version of the repository

AgitatedDove14 yeah it should be..

@<1523701087100473344:profile|SuccessfulKoala55> hey Jake, how do i check how many envs it caches? doing ls -la .clearml/venvs-cache gives me two folders

I'm on Ubuntu 18.04.3

yes

yeah, that's fair enough. is it possible to assign cpu cores? I wasn't aware

i expected to see 2 tasks running, and then when completed the remaining 2 could start. Is this not the expected behavior?

Ok gotchu. I'll do that as soon as I can.

yes, docker compose

yep, setting it to -1 is still caching envs..

yes, I just ran steps 6-12 again from https://allegro.ai/docs/deploying_trains/trains_server_linux_mac/

ok, thanks Jake!

Yeah, I simply used a different port but I got this output:
` (prediction_module) emilio@unicorn:~/clearml-serving$ docker run -v ~/clearml.conf:/root/clearml.conf -p 9501:9501 -e CLEARML_SERVING_TASK_ID=7ce187d2218048e68fc594fa49db0051 -e CLEARML_SERVING_POLL_FREQ=5 clearml-serving-inference:latest
CLEARML_SERVING_TASK_ID=7ce187d2218048e68fc594fa49db0051
CLEARML_SERVING_PORT=
CLEARML_USE_GUNICORN=
EXTRA_PYTHON_PACKAGES=
CLEARML_SERVING_NUM_PROCESS=
CLEARML_SERVING_POLL_FREQ=5
CLEARML_DEFAULT...

I can do curl http://localhost:8080 but it's a remote server so unless I do X forwarding I can't browse it

I am tagging AgitatedDove14 since I sort of need an answer asap...!

unicorn is an alias for the ip of the machine the server is hosted on