inexplicable behavior during deployment, about 19 out of 20 fail #1821
Replies: 3 comments 8 replies
-
|
Same here. Seems like Hetzner is running out of servers... I have opened a support request and just got a response (translated):
In my kube.tf file I have not specified a placement group. I think the code in placement_groups.tf will generate a placement group per nodepool, which means that each virtual server of one nodepool will (try to) be created on a different host. Hosts with available servers seem to be limited, which may result in the failing deployment. I'll try to find a way to skip setting placement groups later, maybe someone already has some input on that? |
Beta Was this translation helpful? Give feedback.
-
|
Same here. Any solutions? |
Beta Was this translation helpful? Give feedback.
-
|
Unfortunately no, no new information and no success :( |
Beta Was this translation helpful? Give feedback.
-
|
That would be great. I have just found a very annoying workaround. I have set the number of agent nodes to 0 and defined 3 control planes. When I run apply, one server is created and the other two disappear. If I then run apply again, another server is created and the other one disappears. After the third time apply I have 3 control plane servers in my Hetzner Console and the terraform script runs successfully... I was then able to do the same with the agent nodes. However, this is not really practicable and in practice my clusters are created via git and a pipeline, where this way is not possible. I find it interesting that it works this way, but creating all servers at once does not work. |
Beta Was this translation helpful? Give feedback.
-
|
I already spoke with Hetzner support last week. They said there was currently an issue with placement groups and that they would inform me about my ticket in writing, which unfortunately never happened. I called again today and was fobbed off rather rudely and told to submit a new ticket. Although there seem to be various issues with creating and even deleting resources, they don't seem to want to make any status or incidents public. I find this level of customer service very disappointing. |
Beta Was this translation helpful? Give feedback.
-
|
oh no that's unfortunate. So far I've been a huge fan of Hetzner and even their support without complaints. I hope this will be resolved soon - since alternatives are rare or 10x more expensive. |
Beta Was this translation helpful? Give feedback.
-
|
Up until now, I've been very happy with Hetzner. But the way things have been going over the last few weeks, it's completely unacceptable. |
Beta Was this translation helpful? Give feedback.
-
|
I used to be a big fan of Hetzner too. The price/performance ratio is unbeatable. We want to run our new cloud suite, which my company is currently developing, on Hetzner, which is currently a huge headache for us. |
Beta Was this translation helpful? Give feedback.
-
|
Got another response from support, saying they improved something in the placement algorithms. I could now create 4 of 6 servers instead of 2 of 6 :) That means I could at least successfully set up the control nodes and can proceed from there. I'm currently experimenting with one other thing, that is disabling parallel requests with Maybe this is also related to limited resources https://status.hetzner.com/de/incident/aa5ce33b-faa5-4fd0-9782-fde43cd270cf |
Beta Was this translation helpful? Give feedback.
-
|
Nice :D I just tried to deploy all servers in Ashburn with the US-East network zone without placement groups, since the site isn't affected according to this incident. Even that fails. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hello everyone,
I'm experiencing some strange behavior, and I'm not sure what's causing it. Most deployments with my kube.tf fail. In the Hetzner console, I can see machines popping up briefly and then disappearing again. I'm not sure if it's a Hetzner Cloud issue or a configuration problem on my part.
I've been trying for days, and it keeps failing with the same behavior.
Is the problem reproducible, or is it just me?
regards,
Erik
Here's the output from terraform apply (after plan):
kube.tf (without comments and spaces):
Beta Was this translation helpful? Give feedback.
All reactions