Chris Lu

There are no actions here that would cause deleting non-empty volumes. The filer connectivity is not possibly related to it either. ``` lock ec.rebuild -force ec.balance -force volume.deleteEmpty -quietFor 24h...

it would purge `.uploads/` under any bucket. These are temporary files generated during uploads. Is it possible to reproduce this?

The logs come from this line: https://github.com/seaweedfs/seaweedfs/blob/master/weed/shell/command_s3_clean_uploads.go#L92 To simplify the test, you can try a similar command like this and see whether it can cause data loss: ``` curl -X...

Please help to verify the fix.

please share reproducing steps and logs.

Could not reproduce on my mac. Can you try to run `weed mount` directly?

Does restarting the mount help? To reproduce this, what file operations you did on the mount?

Try to ec decode the volume

the volume data looks fine. Maybe the filer metadata is out of sync. Can you do this to check the filer metadata: ``` $ weed shell -master=... -filer=... > cd...

does the `sequencer_snowflake_id` have different values in the `master.toml`?