Why doesn't replacing a worker node create a worker node?
Troubleshoot worker node autorebalancing issues.
Virtual Private Cloud Classic infrastructure Satellite
When you replace a worker node or update a VPC worker node, a worker node is not automatically added back to your cluster.
By default, your worker pools are set to automatically rebalance when you replace a worker node. However, you might have disabled automatic rebalancing by manually removing a worker node, such as in the following scenario.
- You have a worker pool that automatically rebalances by default.
- You have a troublesome worker node in the worker pool that you removed individually, such as with the
ibmcloud oc worker rmcommand. - Now, automatic rebalancing is disabled for your worker pool, and is not reset unless you try to rebalance or resize the worker pool.
- You try to maintain a worker node using one of the following commands, but no additional worker node is created to replace it.
- Classic bare metal workers:
ibmcloud oc worker reloadreloads the worker in place (no deletion or addition).ibmcloud oc worker updatemoves the worker to the master BOM version. - VPC bare metal workers:
ibmcloud oc worker reloadreloads the worker in place (no deletion or addition).ibmcloud oc worker replaceoribmcloud oc worker replace --updateare also supported and delete and reprovision the node with a new IP address. - VPC virtual server instance (VSI) workers:
ibmcloud oc worker replace --updatedeletes the existing worker and provisions a new one at the master version.ibmcloud oc worker replacedoes the same at the current patch version. Note thatibmcloud oc worker updateis not supported for VPC worker nodes.
- Classic bare metal workers:
You might also have issued the remove command shortly after the replace command. If the remove command is processed before the replace command, the worker pool automatic rebalancing is still
disabled, so your worker node is not replaced.
To enable automatic rebalancing, rebalance or resize your worker pool. Now, when you replace a worker node, another worker node is created for you.