data_cli generate_snapshot issues

[LEOLIU0127]

I have encountered several questions regarding the generate_snapshot command from the data_cli tool and would like to get some insights on this

1. I kept seeing the E20231020 15:36:15.855079 24 metrics_recorder.cc:160] Unable to cast meter provider from nostd::shared_ptr<MeterProvider> issue running the generate_snapshot with both local/aws platform build. wondering if theres any insights on this particular error
2. Regardless of the above error msg, local platform was still able to generate the snapshot as expected; however the aws version was not
3. the --data_dir is meant to be the BUCKET name, is there a way to config the prefix as well ?
4. for the aws version - is the snapshot supposed to be saved in the --data_dir it specified. in this case, inside the bucket directly ?

[peiwenhu]

Hi,

The error message is related to exporting metrics which is more meaningful for the actual server. Not much for the CLI (at least if the CLI is used off production). If it doesn't impact the process's functioning it should be fine. Although we'll work to address that error.

Is it ok if you use the local version to generate a snapshot file and use aws cli to copy it over to S3? I think we can fix the problem stated here but it makes me wonder if this functionality from the data cli brings more value added than confusion and maintenance. because over time there may be more issues like this and we also need to think about feature parity for other cloud platforms. So maybe the recommendation would be to always treat the data_cli as cloud-agnostic and focused on data manipulation (the data format is all cloud-agnostic).