From 947a705016e7f5fc7d7b173674facecc0acedb15 Mon Sep 17 00:00:00 2001 From: Ruben Hensen Date: Fri, 23 Oct 2026 23:39:49 +0200 Subject: [PATCH] longhorn: poll the backup store every 30 minutes Each poll updates the backupvolume objects, which made up about 8% of the k8s-dqlite writes at the 300s interval. The recurring backup runs once a night, so 30 minutes is plenty. --- apps/templates/longhorn-helm.yaml | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/apps/templates/longhorn-helm.yaml b/apps/templates/longhorn-helm.yaml index 73110a3..578e12e 100644 --- a/apps/templates/longhorn-helm.yaml +++ b/apps/templates/longhorn-helm.yaml @@ -47,7 +47,9 @@ spec: # the backup-target setting. defaultBackupStore: backupTarget: nfs://192.168.1.99:/volume1/microk8s-nfs - pollInterval: 300 + # Every poll rewrites the backupvolume objects, about 8% of all + # datastore writes at 300s. Backups only run nightly. + pollInterval: 1800 defaultSettings: # Move volumes to the new engine image after each Longhorn # upgrade, one at a time. At 0 they stayed behind on 1.7.1/1.7.2.