Files
k7/docs/runbooks/K7StoragePoolFilling.md
G 3c80778156 Release 0.5.0
Observability, usage metering, tenants and quotas, and roles with OIDC login. A Helm chart installs K7 on a cluster you already run, API keys carry a role, and k7-agent no longer needs host privileges.
2026-10-18 23:21:48 +02:00

1.1 KiB

K7StoragePoolFilling

A node's storage pool has been above its threshold for 10 minutes:

Pool Threshold (chart value)
thinpool_data 80% (thinpoolDataRatio)
thinpool_metadata 70% (thinpoolMetadataRatio)
k7d_disks 85% (k7dDisksRatio)

A full thin pool stops every kata-firecracker-devmapper sandbox on the node, and a full k7d disk pool fails k7d creates and forks.

First three commands

ssh <node> lvs -o lv_name,lv_size,data_percent,metadata_percent
ssh <node> df -h /var/lib/k7d/disks
k7 list

Reading them

  • Data filling: delete sandboxes you no longer need (k7 delete), or remove their snapshots (k7 snapshot delete). Snapshot GC removes expired ones every 10 minutes; see K7SnapshotGCFailing if it does not.
  • Metadata filling while data is low: many small thin volumes. Extend the metadata LV (lvextend --poolmetadatasize). It is the one manual storage action the playbook does not take for you.
  • k7d disks: each fork shares its parent's disk until written, so long-lived forks of a busy parent grow. Delete idle forks first.