From 5d46d726e28a1885b5583b76b16f6cdedca57034 Mon Sep 17 00:00:00 2001 From: Michael Date: Sat, 21 Jul 2018 01:14:15 +0200 Subject: [PATCH] Note on missing/incorrect labels and scheduler behaviour with pod anti-affinity (#8973) --- content/en/docs/concepts/configuration/assign-pod-node.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/content/en/docs/concepts/configuration/assign-pod-node.md b/content/en/docs/concepts/configuration/assign-pod-node.md index 7bca696bb5..ce6be282eb 100644 --- a/content/en/docs/concepts/configuration/assign-pod-node.md +++ b/content/en/docs/concepts/configuration/assign-pod-node.md @@ -178,6 +178,8 @@ in the section [Interlude: built-in node labels](#interlude-built-in-node-labels processing which can slow down scheduling in large clusters significantly. We do not recommend using them in clusters larger than several hundred nodes. +**Note:** Pod anti-affinity requires nodes to be consistently labelled, i.e. every node in the cluster must have an appropriate label matching `topologyKey`. If some or all nodes are missing the speficied `topologyKey` label, it can lead to unintended behavior. + As with node affinity, there are currently two types of pod affinity and anti-affinity, called `requiredDuringSchedulingIgnoredDuringExecution` and `preferredDuringSchedulingIgnoredDuringExecution` which denote "hard" vs. "soft" requirements. See the description in the node affinity section earlier.