Closed aronchick closed 3 minutes ago
This PR #4488 mitigates most of the mentioned concerns:
Prints in red (why?)
No longer printing errors in red. Just an Error prefix in red
Doesn't wrap properly
Fixed
The animated fish are still not showing
Brought back the animated fix, but not exactly how it use to be. Your feedback is needed here
It's passing through raw Go errors and not explaining what's going on
That is more on the errors reported from the servers. We did some work on improving the returned errors along with hints, but obviously there is still room for improvement here
It says "4 runs on 2 nodes" - i don't understand this? It's one job submission, one run.
No longer printed that by default. There is already a --node-details
flag that prints a summary of executions across all nodes, and we used to print the summary if there are failures even if --node-details
wasn't set. This is a bit confusing and not needed any more as we are printing more detailed information in the progress tracker anyways
It's not a "not enough nodes to run a job" error, it's a bad image. Saying it's not enough nodes to run a job is backwards.
I agree. This is a bigger topic to solve which is what is the error at the job level if there are multiple executions and each failed for a different reason. Right now the scheduler keeps retrying until it cannot find any more nodes to retry on, and then prints that error message, which is not great. Hopefully it will be improved with https://github.com/bacalhau-project/bacalhau/issues/4015. In the meantime, I did a workaround to just filter out that message on the client side and just print execution level errors
Issues: