Hi MortifiedDove27 ,

do you use output_uri in your Task.init call?

Can you check the files server in the running task?

from clearml import StorageManager print(StorageManager.get_files_server())

No, even new started experiment is still creating images with 172.,
` cat ~/clearml.conf

ClearML SDK configuration file

api {
# Notice: 'host' is the api server (default port 8008), not the web server.
api_server:
web_server:
files_server: `

TimelyPenguin76 Thank you for posting this. I just realized that I changed wrong config. I changed the one on server, but I needed to change the one inside the docker container. Now all works. Thanks for help!

Do you mean the URLs of the previously reported images? These are fixed and will not change, the configuration change will only apply to newly reported images (and you don't have to restart the server 🙂 )

Thank you. I've changed clearml.conf, but url are remain with old ip. Do I need to restart ClearML or run any command to apply config changes?

👍

Hi MortifiedDove27 ,
The image URLs are actually encoded by the ClearML SDK when uploading the images, so your issue is probably the clearml.conf configuration file - when the ClearML SDK rune, it takes the files_server value and uses it to both upload the images, and create the URLs (which are later used by the UI).
I assume in your case, the ClearML SDK is running with files_server: http://172.x.x.x:8081 so the URLs that you get use this value, while the file server is exposed externally with https://25.x.x.x which is why you can get there when manually replacing the links

The solution is to make sure the ClearML SDK (using clearml.conf ) uses the actualy externally exposed address for the files_server so that registered images URL is constructed in a way that will allow externally-run UI to view the images.