Skip to main content
These partitioning recommendations draw from the Defense Information Systems Agency (DISA) Security Technical Implementation Guide (STIG) library. Poolside does not provide STIG-specific implementation guidance or support for customer-defined STIG policies.

Partitioning overview

This page shows recommended partition sizing for single-node Poolside on-premises model deployments running RKE2 and explains how model checkpoints affect storage requirements. The checkpoint estimates cover these model artifacts:
ModelCheckpoint size (GB)
laguna_m_fp8_fp8kv_re_06_2026214
laguna_xs_fp8_fp8kv_re_06_202633
point_v2_04_202625

STIG-compliant partition recommendations

For STIG-hardened Linux nodes, Poolside recommends the following partition layout and sizing as a minimum for the listed models:
MountSizeFilesystemMount optionsPurpose
/boot/efi600Mfat32umask=0077UEFI system partition
/boot1Gxfsnodev,nosuidKernel, initramfs
/50GxfsdefaultsOS root
/tmp2Gxfsnodev,nosuid,noexecTemporary files
/var5Gxfsnodev,nosuidCache, spool, backups
/var/tmp10Gxfsnodev,nosuid,noexecTemporary files
/var/log50Gxfsnodev,nosuid,noexecSystem + audit logs
/var/lib400Gxfsnodev,nosuidkubelet, rancher, containers
/home50Gxfsnodev,nosuidTerraform working directories
/opt/poolside350Gxfsnodev,nosuidModel artifact storage
swap4GswapnoneSwap space
The total recommended disk space for the listed model artifacts is 1 TB.

Dedicated /opt/poolside partition for model artifacts

STIG controls recommend dedicated partitions to prevent system partitions from filling up and causing denial of service. Poolside recommends placing model artifact storage on a dedicated /opt/poolside partition rather than the root filesystem. Without separation, model artifact growth can render the OS unusable.

Inference storage impact

When you run inference workloads, the workloads load model checkpoints from SeaweedFS into storage-backed emptyDir volumes under /var/lib/kubelet. This means you store two copies of the model checkpoint on disk: one copy in SeaweedFS and one copy in the emptyDir volume. Each inference workload consumes disk space equal to its model checkpoint size:
ModelCheckpoint size (GB)emptyDir consumption (GB)Total disk usage (GB)
laguna_m_fp8_fp8kv_re_06_2026214214428
laguna_xs_fp8_fp8kv_re_06_2026333366
point_v2_04_2026252550
Plan for sufficient free space in /var/lib to accommodate your largest model’s checkpoint size when running inference.

Scaling prioritization guidance

If you need to scale storage, allocate additional space in this priority order:
  1. /var/lib: Scales per model deployed in-cluster
  2. /opt/poolside: Scales with model checkpoints
  3. /home: Scales with Terraform files. The Poolside Terraform bundle packages container upgrade images.
Core OS partitions (/, /boot, /tmp, /var, /var/tmp, /var/log) can remain fixed regardless of total disk size.

STIG mount option recommendations

  • /tmp, /var/tmp, and /var/log should have noexec set in the mount options to prevent code from running in temp directories.
  • /var/lib cannot have noexec. The container runtime runs binaries from overlay layers stored there. Adding noexec here breaks Kubernetes pods.