Skip to main content

How to Use HPS in Japan Region (DDN)

Initial Setup​

To begin using File Storage - High Performance (HPS), contact the FPT Support team. After supplying the required info, your account will be activated and a quota assigned.

Required Information

#ItemUnitRequiredDescription
1Tenant/Region—YesThe Tenant and Region where you want to enable the HPS service.
2Storage QuotaGBYesThe amount of storage capacity you plan to use.
3SubnetSubnet CIDRYesProvide the list of subnets needing access so FPT support can configure them.
4PerformanceIOPSNoDesired IOPS performance (default limit applied if not provided).
5BandwidthGbpsNoDesired bandwidth (default limit applied if not provided).

Use with Metal Cloud (a Bare Metal GPU server)​

note

Each subnet has its own endpoint for connecting to the file storage system. Always verify the endpoint and configuration details before performing the mount operation to avoid setup errors.

Use FPT Images​

If using an image provided by FPT, the DDN client is pre-installed. Follow the steps below to mount the File Storage.

Step 1: Create a Mount Point

mkdir /mnt/hps

Step 2: Mount the File System manually

mount -t lustre 100.69.252.1@o2ib,100.69.252.2@o2ib:100.69.252.3@o2ib,100.69.252.4@o2ib:100.69.252.5@o2ib,100.69.252.6@o2ib:100.69.252.7@o2ib,100.69.252.8@o2ib:/scratch /mnt/hps

Step 3: Auto-Mount Configuration

To automatically mount the Lustre file system after reboot, add the following entry to /etc/fstab:

## LustreFS ##
100.69.252.1@o2ib,100.69.252.2@o2ib:100.69.252.3@o2ib,100.69.252.4@o2ib:100.69.252.5@o2ib,100.69.252.6@o2ib:100.69.252.7@o2ib,100.69.252.8@o2ib:/scratch /mnt/hps lustre flock,defaults 0 0

Explanation:

  1. The addresses 100.69.252.1–100.69.252.8 are fixed endpoint addresses of the Lustre servers. /scratch is the filesystem path on the Lustre server.
  2. /mnt/hps is the mount point on the client (created in Step 1).
note

Run mount -a to verify that the mount configuration has no errors.

Step 4: Verify

df -h -t lustre

Expected output shows the filesystem mounted at /mnt/hps with approximately 11T size.

Using Custom Images (Ubuntu)​

If using a custom image running Ubuntu, you must manually build the required packages by following these steps.

Step 1: Build Lustre Client

This step builds the Lustre Client package from source code when a prebuilt package compatible with your kernel version isn't available. Skip this step if a compatible package already exists on your system or if you've previously built the package for the same kernel version.

wget https://s3-sgn10.fptcloud.com/file-storage-resource/exa-client-6.3.4.tar.gz
tar xzvf exa-client-6.3.4.tar.gz
cd exa-client
./exa_client_deploy.py

If your Ubuntu OS is running kernel version 5.15.0-130-generic, you can skip the build and use prebuilt packages:

wget https://s3-sgn10.fptcloud.com/package-pub/lustre-dev_2.14.0-ddn184-1_amd64.deb
wget https://s3-sgn10.fptcloud.com/package-pub/lustre-client-modules-5.15.0-130-generic_2.14.0-ddn184-1_amd64.deb
wget https://s3-sgn10.fptcloud.com/package-pub/lustre-client-utils_2.14.0-ddn184-1_amd64.deb

Step 2: Install Lustre Client

Install the Lustre Client using .deb packages built in Step 1 or downloaded above. Only do this if matching .deb packages are available; otherwise return to Step 1. Kernel version mismatches may cause errors requiring a rebuild.

sudo dpkg -i lustre-dev_2.14.0-ddn184-1_amd64.deb \
             lustre-client-modules-5.15.0-130-generic_2.14.0-ddn184-1_amd64.deb \
             lustre-client-utils_2.14.0-ddn184-1_amd64.deb

Step 3: Configure the Lustre Client

Apply these configuration settings to optimize Lustre Client performance.

Edit /etc/modprobe.d/lustre.conf:

options lnet networks=o2ib(bond0)
options ko2iblnd peer_credits=32 peer_credits_hiw=16 concurrent_sends=64

Edit /etc/lustre_client:

IF1=bond0
IF2=none
IF3=none
IF4=none
IF5=none
IF6=none
IF7=none
IF8=none

Step 4: Create a mountpoint in the OS

mkdir /mnt/hps

Step 5: Mount the File System manually

mount -t lustre 100.69.252.1@o2ib,100.69.252.2@o2ib:100.69.252.3@o2ib,100.69.252.4@o2ib:100.69.252.5@o2ib,100.69.252.6@o2ib:100.69.252.7@o2ib,100.69.252.8@o2ib:/scratch /mnt/hps

Step 6: Auto-Mount Configuration

To automatically mount after reboot, add the following entry to /etc/fstab:

##LustreFS##
100.69.252.1@o2ib,100.69.252.2@o2ib:100.69.252.3@o2ib,100.69.252.4@o2ib:100.69.252.5@o2ib,100.69.252.6@o2ib:100.69.252.7@o2ib,100.69.252.8@o2ib:/scratch /mnt/hps lustre flock,defaults 0 0

The addresses 100.69.252.1–100.69.252.8 are fixed Lustre server endpoints; /scratch is the server filesystem path. /mnt/hps is the client mount point from Step 4.

note

Run mount -a to verify the mount configuration has no errors.

Step 7: Verify

df -h -t lustre

Expected output shows the filesystem mounted at /mnt/hps with approximately 11T size.

Use with a GPU Virtual Machine​

Step 1: Attach Network Interface and Configure IP Address

To connect to HPS Storage for H200, use the pre-created subnet named vm-gpu-hps-net-xxx located under Network > Subnets.

To attach the Network Interface:

  1. Navigate to Instance Management
  2. Select your target VM GPU
  3. Open the Network Interface tab
  4. Click Create NIC
  5. Select the network named vm-gpu-hps-net-xxx

After attaching the interface, access your server and configure the network settings manually if DHCP is not enabled for that interface.

Step 2: Build and Install Lustre Client

This step builds and installs the Lustre Client package from source code.

wget https://s3-sgn10.fptcloud.com/file-storage-resource/exa-client-6.3.4.tar.gz
tar xzvf exa-client-6.3.4.tar.gz
cd exa-client
./exa_client_deploy.py

Step 3: Configure the Lustre Client

Apply these configuration settings to optimize Lustre Client performance.

Edit /etc/modprobe.d/lustre.conf:

options lnet networks=tcp(eth1)
options ko2iblnd peer_credits=32 peer_credits_hiw=16 concurrent_sends=64

Edit /etc/lustre_client:

IF1=eth1
IF2=none
IF3=none
IF4=none
IF5=none
IF6=none
IF7=none
IF8=none

Step 4: Create a mountpoint in the OS

mkdir /mnt/hps

Step 5: Mount the File System manually

mount -t lustre 100.69.252.1@tcp,100.69.252.2@tcp:100.69.252.3@tcp,100.69.252.4@tcp:100.69.252.5@tcp,100.69.252.6@tcp:100.69.252.7@tcp,100.69.252.8@tcp:/scratch /mnt/hps

Step 6: Auto-Mount Configuration

To automatically mount after reboot, add the following entry to /etc/fstab:

##LustreFS##
100.69.252.1@tcp,100.69.252.2@tcp:100.69.252.3@tcp,100.69.252.4@tcp:100.69.252.5@tcp,100.69.252.6@tcp:100.69.252.7@tcp,100.69.252.8@tcp:/scratch /mnt/hps lustre flock,defaults 0 0

The addresses 100.69.252.1–100.69.252.8 are fixed Lustre server endpoints; /scratch is the server filesystem path. /mnt/hps is the client mount point from Step 4.

note

Run mount -a to verify the mount configuration has no errors.

Step 7: Verify

df -h -t lustre

Expected output shows the filesystem mounted at /mnt/hps with approximately 11T size.