Snapper space-aware cleanup on openSUSE: enable qgroups or keep quotas off?

The Decision: Performance vs. Disk Visibility

Enable Btrfs quota groups (qgroups) only if your root partition is frequently near capacity and you require Snapper to prune based on disk usage rather than count. For most users where count-based pruning (e.g., 'keep the last 10 snapshots') is sufficient, keep quotas off to avoid metadata overhead and potential latency.

When is qgroup overhead noticeable?

The overhead of qgroups is not a constant tax but scales with the complexity of operations. On modern kernels (5.15+), the performance hit is significantly mitigated, but issues typically emerge under these specific conditions:

• High Snapshot Count: Once you exceed 500–1,000 snapshots, the time required to calculate shared extents during snapper list can become noticeably slow.
• High Churn: On filesystems with thousands of small files (like databases or large node_modules), the constant metadata updates for qgroups can lead to I/O wait spikes.
• Large Filesystem Size: On multi-terabyte volumes, the initial accounting of qgroups after running snapper setup-quota can take hours and consume significant CPU.

Count-based vs. Space-aware pruning

If your snapshots are primarily generated by zypper transactions, **count-based cleanup is usually sufficient**. These transactions are predictable in size. Space-aware pruning is worth enabling only if:

• You perform large-scale data operations (e.g., downloading 50GB datasets or compiling large projects) that could fill the disk before the count-based cleanup triggers.
• You are running on a constrained VPS environment where a single "heavy" snapshot could cause disk-full errors regardless of how many total snapshots exist.

Timeline snapshots and Quotas

You can safely enable timeline snapshots and quotas together. They operate on different logic layers: timeline rules define when a snapshot is created and deleted, while qgroups define how the system calculates the space consumed. There is no technical conflict, though enabling both will increase the total metadata load mentioned above.

Verification and Implementation

If you decide to enable quotas, use these commands to verify the system is behaving correctly:

# Enable quota groups for the root subvolume
sapper setup-quota root

# Verify Snapper can see used space
sapper list --space

# Check Btrfs qgroup status directly
btrfs qgroup show -/

Note: These commands assume a standard openSUSE Leap/Tumbleweed layout. If you are using MicroOS or a custom subvolume hierarchy, verify your subvolume path in snapper config.