From 1e4087575917320dc1bc2f0581c4df89600a215c Mon Sep 17 00:00:00 2001 From: Tim Wilfong Date: Wed, 15 Nov 2023 14:50:21 -0800 Subject: [PATCH 1/2] Change data_home default to avoid Flexvolume plugin conflict in GKE The current "/home/kubernetes/flexvolume/agent-k8s" default is a directory that the GKE distribution of Kubernetes uses as its Flexvolume plugin directory. https://github.com/kubernetes/community/blob/master/contributors/devel/sig-storage/flexvolume.md --- charts/agent-k8s/values.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/charts/agent-k8s/values.yaml b/charts/agent-k8s/values.yaml index 69562dd..9283244 100644 --- a/charts/agent-k8s/values.yaml +++ b/charts/agent-k8s/values.yaml @@ -28,7 +28,7 @@ agent: # -- Determines if the agent should automatically disconnect from the Scalr agent pool when the service is stopping. disconnect_on_stop: true # -- The agent working directory on the cluster host node. - data_home: "/home/kubernetes/flexvolume/agent-k8s" + data_home: "/home/scalr/agent-k8s" # -- The container task's (e.g., Kubernetes Pod) scheduling timeout in seconds. # The task will be waiting for the scheduling in the queued status; if the cluster # does not allocate resources for the container in that timeout, the task will be switched to the errored status. From 06cf515acae1bedc9227a51dd73327d2a11fc88f Mon Sep 17 00:00:00 2001 From: Tim Wilfong Date: Fri, 17 Nov 2023 13:38:24 -0800 Subject: [PATCH 2/2] Change data_home to /home/kubernetes/bin/scalr/agent-k8s Unlike "/home", the "/home/kubernetes/bin" directory should not be mounted with noexec, so this should work. This is tested with GKE, but not with other K8s providers. --- charts/agent-k8s/values.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/charts/agent-k8s/values.yaml b/charts/agent-k8s/values.yaml index 9283244..509f0f1 100644 --- a/charts/agent-k8s/values.yaml +++ b/charts/agent-k8s/values.yaml @@ -28,7 +28,7 @@ agent: # -- Determines if the agent should automatically disconnect from the Scalr agent pool when the service is stopping. disconnect_on_stop: true # -- The agent working directory on the cluster host node. - data_home: "/home/scalr/agent-k8s" + data_home: "/home/kubernetes/bin/scalr/agent-k8s" # -- The container task's (e.g., Kubernetes Pod) scheduling timeout in seconds. # The task will be waiting for the scheduling in the queued status; if the cluster # does not allocate resources for the container in that timeout, the task will be switched to the errored status.