Skip to content

S Drive (Shared Lab Storage)

The lab S Drive is persistent shared storage hosted by Monash eSolutions. Unlike M3 project storage (500 GB shared quota), the S Drive is backed up and does not get purged.
UNC path \\ad.monash.edu\shared\ENG-ECSE\Robotics-M3
Windows path S:\ENG-ECSE\Robotics-M3
Purpose Archive completed training runs, share assets across the lab
Access Request via Lingheng — Monash account required

Warning

S Drive uses SMB (network file share) — it is not suitable for training I/O. Use it only for one-time archiving and restoring. Active training must run on M3 project storage or scratch.

Members

Access group: Robotics-M3 | 33 members | Last updated: 2026-05-03

Name Role
Dana Kulić PI
Lingheng Meng Admin — drive manager
Kavi Katuwandeniya Research Scientist
Somayeh Hussaini Visiting Post-doc
Rhys Newbury Post-doc
Shiyu Xu Post-doc/PhD
Lirui Guo PhD Student
Phan Le PhD Student
Malak Sayour Research Student
Mohammad Zawari Research Student
Alana Hayes Chen Member
Aron Thomas FYP Student
Ben McShanag FYP Student
Blake Matheson FYP Student
Duleesha Gunaratne FYP Student
Isabella Moffat FYP Student
Jack Bastasin FYP Student
Jack Thornborrow FYP Student
Jacinta Yao FYP Student
Jeremia Yovinus FYP Student
Jesse Kipchumba FYP Student
Jordan Rozario FYP Student
Lucas Cooper FYP Student
Maxwell Fergie FYP Student
Michael Fitzgerald FYP Student
Omar Salem FYP Student
Raphael Schwalb FYP Student
Suhaas Kataria FYP Student
Taine Fryer FYP Student
Tavishi Saxena FYP Student
Tianyi Xiao FYP Student
Vedansh Malhan FYP Student
William Ngo FYP Student

Getting Access

Email Lingheng (lingheng.meng1@monash.edu) with your Monash authcate ID or Monash email address. He will add you as a Group Member. Allow up to 3 hours for permissions to propagate.

Folder Structure

The S Drive follows the same two-level convention as M3: project folder → stack_{name}/ subfolder.

S:\ENG-ECSE\Robotics-M3\
  FYP2026S1_3487\
    stack_alice\          ← Alice creates this herself
      checkpoints\
      logs\
    stack_bob\
  PhD_2023_dave\
    stack_dave\
  shared_assets\          ← shared lab assets (Isaac Sim .sif, models)

Project folder naming (same as M3):

Role Format Example
FYP student FYP{YYYY}S{1\|2}_{ProjectID} FYP2026S1_3487
PhD student PhD_{YYYY}_{firstname} PhD_2023_dave
Postdoc Postdoc_{YYYY}_{firstname} Postdoc_2026_somayeh

Note

Supervisors create the project folder — the same rule as M3. Students create only their stack_{name}/ inside it. This ensures consistent structure and makes it possible to manage the drive as people come and go.

Each person is responsible for creating their own stack_{name}/ folder inside the project folder. Do not write into someone else's stack.

Note

There is no per-folder access control — every Group Member can read and write anywhere in the share. This is trust-based, the same as M3 project storage.

Storage Policy

The S Drive is shared lab storage. To keep it manageable:

  1. Follow the folder convention. All data must live under {project_folder}/stack_{name}/. Non-standard folders and loose files at the root will be reorganized by the admin without notice.
  2. Don't leave orphaned data. When you finish a project, update the STATUS file in your project folder (same values as M3: ACTIVE, DONE-{YYYY-MM}, ARCHIVE, OK-TO-DELETE).
  3. Shared assets go in shared_assets/. Large files shared across the lab (Isaac Sim .sif, base models) belong in shared_assets/ — not duplicated in each person's stack.

Mounting on M3

Run on the login node only — not inside a SLURM job.

Step 1 — Mount

gio mount 'smb://MONASH;{your_monash_id}@ad.monash.edu/Shared'

Enter your Monash password when prompted (not your M3/MASSIVE password).

Step 2 — Create a symlink for convenience

ln -s /run/user/$(id -u)/gvfs/smb-share:domain=MONASH,server=ad.monash.edu,share=shared,user={your_monash_id}/ENG-ECSE/Robotics-M3 ~/s_drive
ls ~/s_drive   # should show S Drive contents

Step 3 — Unmount when done

gio mount -u 'smb://MONASH;{your_monash_id}@ad.monash.edu/Shared'

Warning

The mount is session-scoped. Compute nodes cannot see it. Do NOT use S Drive paths inside SLURM job scripts.

Archiving Data from M3 to S Drive

Use this to free M3 project quota by moving completed runs to S Drive.

SHARE=~/s_drive
PROJECT=FYP2026S1_3487       # your project folder name
MY_NAME=alice                 # your stack name

# Archive a training run
rsync -av --progress \
  /fs04/mh42/${PROJECT}/stack_${MY_NAME}/logs/<run-folder>/ \
  $SHARE/${PROJECT}/stack_${MY_NAME}/logs/<run-folder>/

# Verify the copy, then free M3 space
rm -rf /fs04/mh42/${PROJECT}/stack_${MY_NAME}/logs/<run-folder>

Restoring from S Drive back to M3

rsync -av \
  $SHARE/${PROJECT}/stack_${MY_NAME}/logs/<run-folder>/ \
  /fs04/mh42/${PROJECT}/stack_${MY_NAME}/logs/<run-folder>/

S Drive  ←→  (rsync via login node, one-time transfers)  ←→  M3 scratch  ←→  M3 project
(archive)                                                      (active jobs)   (lean, active only)
  1. Mount S Drive on login node
  2. Copy needed data: S Drive → scratch (large, fast Lustre)
  3. Submit SLURM job — reads/writes scratch or project only
  4. After job, archive results: scratch/project → S Drive
  5. Clean up scratch and project to free space

For large shared assets (e.g. Isaac Sim .sif, ~20 GB): keep one copy in s_drive/shared_assets/ and copy to scratch before jobs — do not keep duplicate copies in the project folder.

Mounting on Windows (personal laptop)

Requires Monash VPN connected first.

  1. Open File Explorer → This PC → ⋯ → Map network drive
  2. Drive: S:, Folder: \\ad.monash.edu\shared\ENG-ECSE\Robotics-M3
  3. Tick Reconnect at sign-in and Connect using different credentials
  4. Credentials: MONASH\{your_monash_id} + Monash password

Speed Reference

M3 Lustre (project/scratch) S Drive (SMB)
Sequential read/write 200–500+ MB/s ~50–150 MB/s
Many small files Reasonable Very slow (~1–5 MB/s)
100 GB transfer ~3–8 min ~12–33 min
Suitable for training I/O Yes No

Priority Targets for Freeing M3 Space

  1. Old training checkpoints and logs no longer actively used
  2. Personal apptainer sandboxes (apptainer/sandbox/, ~20 GB each) — rebuild from s_drive/shared_assets/ if needed
  3. Duplicate conda environments — coordinate with teammates to share one copy