backup2nrm

• Last modified: tor aug 12, 2021 09:13
• Sign: JN

Description

Script for transferring UPPMAX files to backup server on NRM.

The current backup server is named nrmdna01.nrm.se and has limited access.

For transfer, we use rsync over ssh.

Assuming file structure similar to:

• On rackham:

    /proj/projectname/folder/to/transfer
  

• On NRM (where XXX is a department acronym, currently in use: BIO, BOT, ZOO, and nrmuser is the user in the Windows ActiveDirectory)

    /projects/XXX-projects/nrmuser
  

Installation

0. From NRM IT-support, get information about
   1. Name and IP address for the NRM backup computer (below called nrmcomputer).
   2. User name on the backup computer (prob. same as your Windows user). Below called nrmuser.
   3. Directory structure on NRM backup computer (e.g. /projects/BIO-projects)
   4. How to set up "passwordless communication between uppmax and nrmdna01.nrm.se"
1. Login to UPPMAX (rackham.uppmax.uu.se)
2. Clone the repository from https://github.com/NBISweden/backup2nrm.git, using
   • git clone https://github.com/NBISweden/backup2nrm.git
3. Put the file backup2nrmdna01 in your PATH (on UPPMAX)
4. For convenience, edit the backup2nrmdna01 file to add values for
   1. $nrmuser
   2. $email
   3. $department
   4. ($nrmpath)
   5. ($cpuaccount)

Usage

The script performs a number of tasks:

1. Gather information on the source (on uppmax) and destination (on nrm) folders
2. Sends mock data to measure the transfer speed and estimate the time needed for complete transfer (optional step)
3. Creates a script with rsync commands for the queue system (SLURM) on uppmax
4. Submits the script to SLURM (optional)

All steps can be run at once. For example (run on uppmax, and with default values edited in the script):

[rackham]$ backup2nrmdna01 -a -s /proj/snic1234-56-789/delivery012345

This will transfer the folder delivery012345 to the destination at NRM, which --- if values for --nrm-user and --department are provided --- is nrmdna01.nrm.se:/projects/XXX-projects/nrmuser/delivery012345.

The full command may be given as (example)

backup2nrmdna01 \
    --department=BIO \
    --nrm-user=joebro \
    --autotime \
    --cpuaccount=snic1234-45-789 \
    [email protected] \
    --submit \
    --verbose \
    /proj/snic1234-56-789/delivery012345

It may, however, be convenient to first generate the slurm script (saved in the current working directory), then manually check - and perhaps modify the entries, and finally submit the slurm-script manually. This may be accomplished like this (again with default values added in the script):

[rackham]$ backup2nrmdna01 -a /proj/snic1234-56-789/delivery012345
[rackham]$ less rsync-12082021.0-delivery012345-snic1234-56-789.slurm
[rackham]$ sbatch --test-only rsync-12082021.0-delivery012345-snic1234-56-789.slurm
[rackham]$ sbatch rsync-12082021.0-delivery012345-snic1234-56-789.slurm

For more details information, use backup2nrmdna01 -h.