GKE Node Pool can't register instance as node in multi zonal

[calvin_sadewa]

I am having problem in provisioning node GKE Node Pool in multi zonal setup (asia-southeast2a, b, c). three instance are created in (one for each zonal), but only asia-southeast2a instance are registered as node, asia-southeast2b and asia-southeast2c don’t.

GKE version is

v1.29.8-gke.1096000

I have tried:

• recreating and deleting node pool through terraform IaC 3x, doesn’t resolve issue.
• try resizing node pool through GCP UI 2x for affected node pool, doesn’t resolve issue
• try creating a new test node-pool on verify the issue, to check, it’s having the same issue instance in asia-southeast2b and asia-southeast2c are having trouble registered as available node
• followed through Node registration troubleshoot guide over https://cloud.google.com/kubernetes-engine/docs/troubleshooting/node-registration?hl=en#node-registration-checker

result from Node registration checker on problematic instance are complete successfully

0m0.001s
[  440.775499] node-registration-checker.sh[2020]: Thu Oct 17 10:05:51 UTC 2024 - ** Here is a summary of the checks performed: **
[  440.775656] node-registration-checker.sh[4561]: ------------------------------
[  440.775723] node-registration-checker.sh[2020]: Service    DNS      Reachable
[  440.775877] node-registration-checker.sh[4563]: ------------------------------
[  440.775928] node-registration-checker.sh[2020]: LOGGING    true     true
[  440.775959] node-registration-checker.sh[2020]: GCS        true     true
[  440.775994] node-registration-checker.sh[2020]: GCR        true     true
[  440.776094] node-registration-checker.sh[2020]: Master     N/A      true
[  440.776210] node-registration-checker.sh[4565]: ------------------------------
[  440.776252] node-registration-checker.sh[2020]: Master     Healthz  Version
[  440.776369] node-registration-checker.sh[4567]: ------------------------------
[  440.776409] node-registration-checker.sh[2020]:            ok       v1.29.8-gke.1096000
[  440.776512] node-registration-checker.sh[4569]: ------------------------------
[  440.776557] node-registration-checker.sh[2020]: Service Account: 74465351471-compute@developer.gserviceaccount.com - enabled: true
[  440.776592] node-registration-checker.sh[2020]: Kubelet logs available: Yes(see above)
[  440.776624] node-registration-checker.sh[2020]: Thu Oct 17 10:05:51 UTC 2024 - ** Completed running Node Registration Checker **

How should i resolve this issue?

[calvin_sadewa]

I have fixed it by upgrading Kubernetes version to v1.30, and now the instance is registered as node normally.

for context i was trying to play around with zonal deployment, consolidating all deployment to zonal a. It seems as old version 1.29 was in deprecation, and after all node & instance vacated zonal b or c, GKE could no longer register new instance in zonal b or zonal c as node.