Hi Andor,
Globus will ignore symlinks discovered as part of a recursive transfer.
When part of the submitted paths (entries that were selected in the
web interface), Globus will treat symlinks as normal files/directories,
copying the data they point to.
More info here:
https://docs.globus.org/faq/transfer-sharing/#how_does_globus_handle_symlinks
How did you calculate the size of the data set? I ask because, while
the Globus behavior will be different than that rsync command, it should
not have copied less data than the rsync command. I believe the
difference is only that rsync will have created the symlinks on the
destination, and Globus will not have. In the rsync case, the
additional symlinks should not increase the size of the dataset (beyond
the negligible size of the symlink itself).
Mike
On 9/21/2021 7:06 AM, Andor Kiss wrote:
>
> Hello,
>
> I copied a genome assembly project (12 TB) from my local HPC cluster
> to our lab workstation using RSYNC - to ensure that all my symlinked
> files and directories withiin the 12 TB were preserved.
>
> When I tried to copy this "backup" from my workstation to Ohio
> Supercomputer Center, I was recommended to try GLOBUS. Using the GUI, I
> tried the transfer, but only 7 TB was moved. All the symlinked
> files/directories were /not /transferred.