bulk transfer

From Cheaha
Jump to navigation Jump to search

Attention: Research Computing Documentation has Moved

Please use the new documentation url for all Research Computing documentation needs.

As a result of this move, we have deprecated use of this wiki for documentation. We are providing read-only access to the content to facilitate migration of bookmarks and to serve as an historical record. All content updates should be made at the new documentation site. The original wiki will not receive further updates.

Thank you,

The Research Computing Team

UAB has an Enterprise contract with, which is currently in BETA.

This page describes what we have learned about doing bulk-transfers of data.

Warning: size limitations claims to have a 5G max filesize limit

  • Now 15G max filesize (2/18/016; private email), but "that file size limit is still considered to be in a beta phase"

There was a rumor this would be increased in early 2016. Offically, this is all we know:

If you need to work around this, you can use the Linux "split" utility

# chop file into 4G pieces

split \
--bytes=4000m \
big_file.fastq.gz \

# record checksums of original and chunks

md5sum \
big_file.fastq.gz \
big_file.fastq.gz.split4g.* \
> big_file.fastq.gz.md5

Warning: time stamps

When using an FTP client to transfer data up, it is easy to lose both modification and creation timestamps. In particular, many clients will (optionally) preserve modification time, but few will (optionally) preserve creation date.

FTP client cost platform preserve mod_date preserve create_date
lftp free linux/cmd_line yes no
SmartFTP $$ Win Only/GUI yes can be enabled
fileZilla free linux & win/GUI_only can be enabled no
ftp_ssl free linux/cmd_line yes no

Filezilla on create times

Warning: Shared-to-you folders can't be moved

If someone creates a folder and shares it to you (as an reader, editor, co-owner, etc), it will live in your top level directory, and you will NOT be able to move it into any subfolder.

"Currently, users can't rearrange their own view of folders they are invited to collaborate within. As you note, when someone invites you to collaborate in a folder that you have never had access to before, you will see that folder on your root level." []

Workaround: if your collaborator makes you the full owner of the folder, then you will be able to move it.

Fix timeframe: "We've heard requests that people be able to rearrange their views before, and this is being considered as part of a larger product experience change next year" []

lftp mirror -R examples (UPload)

lftp mirror

  • "mirror" copies directory hierarchies DOWN from to local
  • "mirror -R" copies directory hierarchies UP from local to

error handling

  • the box server frequently looses connection (fails) on particular files
  • just re-run the "mirror -R" and it will upload only new/failed files..

Interactive lftp


> user users_BOX_external_password
> mirror --parallel=10 -R local_src_dir box_dest_dir

Single-line lftp (non-shared box)

Warning: This one makes the password visible to "ps" so should only be used on personal machines

lftp -u,users_BOX_external_password << EOF

mirror --parallel=10 -R local_src_dir box_dest_dir

scripted lftp

cat > box_upload.lftp << EOF

user users_BOX_external_password
mirror --parallel=10 -R local_src_dir /box_dest_dir
chmod 700 box_upload.lftp
lftp -f box_upload.lftp ; echo lftp_RC=$?

lftp mirror examples (DOWNload)

scripted lftp


  • --loop keep restarting until there are no new files left to download - helps if someone else is uploading to that directory while you're downloading it!
  • -v verbose level 1: includes bytes transferred and transfer speed.
  • --parallel=10 use 10 concurrent TCP/IP connections (much faster)
cat > box_download.lftp << EOF

user users_BOX_external_password
mirror --loop -v --parallel=10 /box_remote_src_dir local_dest_dir
chmod 700 box_download.lftp
lftp -f box_download.lftp ; echo lftp_RC=$?

Issues to resolve

  • Routing over Iternet2
 * we see our traffic randomly going over the commodity internet

Linux Support

Unfortunately, Box doesn't provide a Linux client (is it on the road map?).