Diskusage Explorer

Content of Folders

Warning: At the moment this feature is only available on Béluga and Narval.

You can get a breakdown by folder of how the diskspace is being consumed in your home, scratch and project spaces. That information is currently updated once a day and is stored in an SQLite format for fast access.

Here is a walkthrough of how to explore your disk consumption, using the example of the project space def-professor as the particular directory to investigate.

Ncurse User Interface

Choose a project space you have access to and which you wish to analyze, which for the purpose of this discussion we will assume to be def-professor.

[name@server ~]$ diskusage_explorer /project/def-professor

This command loads a browser that shows the resources consumed by all files under any directory tree

Navigating your project space with duc's ncurse tool

Type c to toggle between consumed disk space and the number of files, q or <esc> to quit and h for help.

If you are only interested in a sub-directory of this project space and do not want to navigate the whole tree in the ncurse user interface,

[name@server ~]$ diskusage_explorer /project/def-professor/subdirectory/

A complete manual page is available with the man duc command.

Graphical User Interface

Note that when the login node is especially busy or if you have an especially large amount of files in you project space, the graphical interface mode can be slow and choppy. For a better experience, you can read the section below to run diskusage_explorer on your own machine.

Note that Compute Canada recommends the use of the standard text-based ncurse mode on our cluster login nodes but diskusage_explorer does also include a nice graphical user interface (GUI).

First make sure that you are connected to the cluster in such a way that SSH is capable of correctly displaying GUI applications. You can then use a graphical interface by means of the command,

[name@server ~]$ duc gui -d /project/.duc_databases/def-professor.sqlite  /project/def-professor

You can navigate the folders with the mouse and still type c to toggle between the size of the files and the number of files.

Navigating your project space with duc's GUI tool

Browse faster on your own machine

First install the diskusage_explorer software on your local machine and then, still on your local machine, download the SQLite file from your cluster and run duc.

rsync -v --progress username@beluga.calculcanada.ca:/project/.duc_databases/def-professor.sqlite  .
duc gui -d ./def-professor.sqlite  /project/def-professor

This immediately leads to a smoother and more satisfying browsing experience.

Space and inode usage per user on cedar

On cedar, it is possible for any member of a group to run diskusage_report with the following options --per_user and --all_users to have a break-down usage per user. The first option gives displays only heavy users. In other terms, members of the group who have more files and/or occupy more space. When both options are used, the command gives the break-down usage for all members of the group. This is a handy command that helps to identify the users within a group who have more files and/or a large amount of data and ask them to better manage their data by reducing their inode usage for example.

[user01@cedar1 ~]$ diskusage_report --per_user --all_users Description Space # of files /home (user user01) 109k/50G 12/500k /scratch (user user01) 4000/20T 1/1000k /project (group user01) 0/2048k 0/1025 /project (group def-professor) 9434G/10T 4062k/5000k Breakdown for project def-professor (Last update: 2023-05-02 01:03:10) User File count Size Location

---

        user01          141567             4.00 GiB              On disk
        user02           11926             3.74 GiB              On disk
        user03          725383          6121.03 GiB              On disk
        user04            4010           377.86 GiB              On disk
        user05         1296466           262.75 GiB              On disk
        user06         1010996            60.51 GiB              On disk
        user07          848066          1721.33 GiB              On disk
        user08           26516           947.23 GiB              On disk
         Total         4064930          9510.43 GiB              On disk

Breakdown for nearline def-professor (Last update: 2023-05-02 01:01:30)

          User      File count                 Size             Location

---

      sbelharb               5          1197.90 GiB     On disk and tape
         Total               5          1197.90 GiB     On disk and tape