Node name collision

Hi

I have a kubernetes setup with some nodes in a autoscaling group. The autoscaling killed several nodes and later relaunch then, but now consul reports this:

2019/11/18 21:48:42 [WARN] consul.fsm: EnsureRegistration failed: failed inserting node: Error while renaming Node ID: “f41dc722-2c99-0430-bc8a-d86f5d14cbed”: Node name ip-172-26-58-159.eu-central-1.compute.internal is reserved by node cfe0d25f-4981-1499-1ee7-2c702de94200 with name ip-172-26-58-159.eu-central-1.compute.internal

From what i understand, the new node is giving some problems because the name it is still in use by the old node, that will never come back. I assume that the old node is in failed state and the new one can’t finish the registration due to the name collision.

I tried the consul force-leave by name and node id, but failed to solve this problem

So what can i do to solve this? The only way i’m seeing is to kill the new node and force-leave the name until it is gone… if that even works

thanks for the help
Daniel

Hi Daniel, which version of Consul are you running? This PR https://github.com/hashicorp/consul/pull/5485 looks to have fixed this in Consul 1.5.1.

i’m using still 1.4, so looks i’m still affected by this! :frowning:

I do plan to upgrade in the next few months, so luckily i will
not be affected by this for a long time. We also plan to change the auto-scaling group host names so this is less a problem.

Thanks!

1 Like