This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(224.00000000000003, 28.000000000000004%, 31%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EMG%3C/text%3E%3C/svg%3E)
Hi everybody,
in an AKS training, we observed the following behavior:
Several participants could not bring up their previously working AKS cluster with 3 nodes in 1 node pool after it has been stopped via the azure portal.
The node pool ist stuck in state "starting" for several hours.
Deleting the cluster took a very long time, but for some it ultimately worked so that they could recreate it.
Any thoughts on how this could happen and how to resolve the situation without deleting and recreating the cluster?
Thanks!
Markus
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(208, 35%, 30%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ESM%3C/text%3E%3C/svg%3E)
Hello @Markus Geiger ,
Thanks for posting your query on Microsoft QnA.
I would like to share a word of caution before answering the question for your specific scenario. It is important that you don't repeatedly start/stop your cluster. Repeatedly starting/stopping your cluster may result in errors. Once your cluster is stopped, you should wait 15-30 minutes before starting it up again.
In your case, I understand that the cluster is stuck is "starting" state for hours when you start the cluster from Portal.
I would recommend you to perform the stop as well as start of your cluster via cli using the below commands. If you still see the same issue persisting, then kindly drop a comment here and tag me so that i can look further into this. Please note, starting and stopping a cluster via cli also takes a few minutes. If the nodepool size is small then it should not take more than 5 minutes (this is what I have observed from experience).
az aks stop --name <cluster-name> --resource-group <rg-name>
az aks start --name <cluster-name> --resource-group <rg-name>
Hope this helps. Looking forward to hear back from you. Please do tag me in your reply. Thank You.
Hi @srbhatta-MSFT , thanks for your response!
We didn't succeed so far, but we got one more clue on why this may be happening:
Seems like the folks who have the problem also have deleted or moved their log analysis workspace and got this error: "The Log Analytics workspace(...) you originally chose to monitor this cluster has been moved or deleted. Please choose a workspace to re-enable monitoring."
Since this problem cannot be resolved in a stopped or failed cluster, we may have kind of a hen-and-egg-problem here
Hi @srbhatta-MSFT , here's an update:
the cluster startup has timed out after some hours and remained in state "Failed (Running)". In this state I was able to stop it.
After a while, using "az aks start --name <cluster-name> --resource-group <rg-name>", I tried to start it again. It is now in state "Starting (Running)" for approx an hour.
Any attempt to stop it using "az aks stop --name <cluster-name> --resource-group <rg-name>" results in
---
(OperationNotAllowed) managed cluster is in Provisioning State(Starting) and Power State(Running) and agent pool agentpool is in a non terminal state, stopping cannot be performed
Code: OperationNotAllowed
Message: managed cluster is in Provisioning State(Starting) and Power State(Running) and agent pool agentpool is in a non terminal state, stopping cannot be performed
---
We leave the cluster in this state until the end of the training, in approx 24 hrs in case you want to look into it. After that, the RG will be deleted.
Hi @Markus Geiger ,
Looking at the above error, it would mean that the cluster hasn't completely started, which is why stopping the cluster is throwing error.
Could you try and update the state of cluster and see if things are then working properly?
az resource update -n <aks cluster name> -g <rgname> --namespace Microsoft.ContainerService --resource-type ManagedClusters
Hi @srbhatta-MSFT ,
az resource update -n <aks cluster name> -g <rgname> --namespace Microsoft.ContainerService --resource-type ManagedClusters
gave me:
(LogAnalyticsWorkspaceNotFound) Unable to retrieve log analytics workspace. Resource ID: .... If resource is intentionally deleted to stop monitoring, please also disable monitoring addon by following doc: https://aka.ms/aks-disable-monitoring-addon. If resource is deleted by mistake, please re-create it by following doc: https://aka.ms/new-log-analytics, and enable new workspace resource id by following doc: https://aka.ms/aks-enable-addons
I followed https://aka.ms/aks-disable-monitoring-addon to disable the monitoring addon after which the cluster started again.
Seems like deleting/moving the log analytics workspace was the culprit.
I think that if this is preventing the node pool from starting up, it should fail fast and with a clear error message.
Thanks for you help!
Markus
@Markus Geiger yes, that's right. This confirms that deleting the log analytics workspace caused this issue to crop up.
If you feel that the information provided by me was helpful and the conversation in this thread was engaging, please consider to accept as answer so that this thread can be refereed by other community members in future. Thank you.