I'm currently trying to copy a largish file (17GB) from one s3 bucket to another which both reside in the same region. I'm running this in a Kubernetes job running with a pod that has up to 10GB of RAM and 2 vCPUs. It appears that rclone only ever uses up 8MB of memory and practically no CPU. I wasn't sure if this was because it was a remote to remote transfer or something involved with my settings.
Ultimately, it doesn't appear to affect the copy speed at all no matter how much I play with transfers, s3-upload-concurrency, or s3-chunk-size. The copy always seems to take around 10-12 minutes.
What is your rclone version (output from rclone version)
rclone v1.52.3
- os/arch: linux/amd64
- go version: go1.14.7
The command you were trying to run (eg rclone copy /tmp remote:tmp)
if you transferring a single file, then flags like --transfers have no effect.
the transfer is limited by internet speeds of the computer running rclone.
so no matter of amount of tweaking can change that hard limit.
what is the internet speeds of the pod?
flags like --s3-upload-concurrency have a limited effect, often not much difference between the default of 4 versus 16