Yes, I am using a self-hosted server.

So we ran some experiments(tasks) and uploaded the artifacts to local storage. We want to move them to a s3 storage

Furthermore we want to move existing datasets on one s3 server to another s3 server

The clearml-server is running on a remote machine and I do port-forwarding via ssh to localhost to access the web interface. Not sure if this is the cause

Yeah, but I still need to update the links in the clearml server

ah nice, thanks a lot 🙂

Thanks SweetBadger76 🙂

I added more information here https://github.com/allegroai/clearml/issues/697

That’s not as straightfoward as it might seem.. the clearml client lib handles artifacts as read only dict. I am not taking a look at changing the values directly in mongodb

no and no

around 100-200

For anyone else struggling with this:
Shell into the mongo container and run
db.task.find({ "project": "<PROJECT_ID>" }, { "execution.artifacts.$": 1 }).forEach(function (doc) { Object.keys(doc.execution.artifacts).forEach(function (k) { doc.execution.artifacts[k].uri = doc.execution.artifacts[k].uri.replace("old_host", "new_host"); db.task.updateOne({ _id: doc._id }, { $set: { execution: doc.execution } }) }) });

1.3.2 as client

Hey SweetBadger76
By using Task.create and specifying the path to the script clearML is able to collect all the source information, however the experiment then is a draft and does not upload anything

Like the relations between the datasets. If I have dataset A as a child of B and B is a child of C I would like to have some textual/visual output like

C<-B<-A