Xcp-ng 8.2 no more recognized with Cloudstack since last update
-
@AlexanderK Just started playing with CS. Trying to see if this might be a better fit for our endusers than the current XOA. XOA is okay, but definitely more focused and defined for admins of the full XCP-ng stack vs. just wanting to deploy VMs. It's possible XO 6 might solve a lot of my issues, but we're looking around for other options that will utilize XCP-ng underneath the covers while maybe giving a better frontend experience for customers.
At this point I'm just trying to get a rather vanilla XCP-ng host in my lab connected to CS so I can learn the architecture better but I'm running into issues. I won't deny that I'm a noob with CS so some of the components are still foreign to me but simply connecting a host is giving me troubles and documentation seems a little spotty. Any ideas? This might need to go into a new thread.
Error:
"Cannot transit agent status with event AgentDisconnected for host....Unable to transition to a new state from Creating via AgentDisconnected"
-
@AlexanderK im basically trying to deploy cs with 8.2.1 where im limited in interfaces, and im on public ip space.
-
@jimmymiller cs vm can communicate with xcp-ng host?
For endusers cs is better solution. It can create networks, vpcs and many more with the push of a button.The difficult part of cs is setting it up, learn how to use it and teach others how to use it.
I using it around 4-5 users. With the use of terraform it is something amazing.When i was setting it up i have found somewhere, apache recommends 3 months of comprehend network infra of cs and then 3 months of debugging.
And...remember this... always check logs. Usually all issues can be solved easily.
-
@dingo try it first at a lab enviroment.
There was a guide somewhere giving info about setting it up with VirtualBox and understand networking.
https://www.shapeblue.com/virtualbox-test-env/Have a look here
-
@AlexanderK dont think you follow, i own, meaning they are my machines, 4 large bare metal hosts on public ip space, each has only 1 nic ..... with xcp-ng installed,
id like to use CS to provision them, but cant get my head around the public / private ip spaces, when all my ips are public on a /26 ... each box does have a second nic, but it is not wired to the switch, only 1 nic is, and they are in a colo, so i have only 1 port per machine. -
@AlexanderK Yes. The hosts and CS Management server are on the same L2/no firewalls. I think it has to do with me wanting to use an "L2 network topology" out the gate but from the reading I'm doing, this is definitely atypical. I've blown away the zone and reconstructed it in what I think is a "normal" CS config now.
I've gotten the hosts attached at least, but now I'm seeing issues with getting the SystemVMs up and running. I did check the 'use local host storage' option in the zone config and I can see CS is at least touching/renamed the local SR.
Secondary Storage Vm creation failure in zone [xxxxxx]. Error details: Unable to allocate capacity on zone [2] due to [null]. Console proxy creation failure. Zone [xxxxxx]. Error details: Unable to orchestrate start VM instance {"id":20,"instanceName":"v-20-VM","type":"ConsoleProxy","uuid":"f138df9e-92c4-4cfd-9333-7d8396c436e6"} due to [Unable to acquire lock on VMTemplateStoragePool: 31].My storage network is completely private and everything is hanging off a single bond0 where I have VLANs carved down to the XCP-ng host, but I haven't configured any networks within XO. I assume CS was supposed to take care of that?
I'm also not seeing the VMs show up in XO. Am I supposed to see instances and SystemVMs in XO?
Thanks for any help.
-
No luck for me trying out CS and XCP-ng in May 2025. I cannot add any hosts.
Still stuck here even after redeploying CS and blowing away the XCP hosts.
"Cannot transit agent status with event AgentDisconnected for host 8, management server id is 206863096530162,Unable to transition to a new state from Creating via AgentDisconnected"
XCP 8.2.1
CS 4.20.0.0 -
@amosgiture
check the cloudstack logs
tail -f /var/log/cloudstack/management/management-server.log
there will be there more info about the issue.always checking logs help pinpoint the issue.
i have many deployments with cs and xcp-ng.
-
Indeed, and also you need to ask CloudStack people, because that's very likely related to CS and not XCP-ng
-
@olivierlambert
Thanks!
Looks like the guys at CS have not really tested or maybe do not have the resources to test CS with all the hypervisors. The fact that XCP 8.2.1 or 8.3 is not listed here may mean that it is not yet tested.
https://docs.cloudstack.apache.org/en/4.20.0.0/releasenotes/compat.html#supported-hypervisor-versions.
XCP 8.2 is near EOL, 8.3 LTS is almost out and, CS 4.21 is out in August 2025 so I'll wait and see. -
@amosgiture
xcp-ng 8.2 and 8.2.1 is working without any issue.
Have you checked the logs?
it is recognized as xenserver 8.2.1
