1

I am trying to setup a cloudstack (v4.4 on CentOS 6.5) management instance to talk to one physical host with XenServer (6.2) on it.

I have got so far as it setting the zone/pod/cluster/host and it can see the XenServer machine. Primary storage is also visible to it - I can see it in the dashboard. However it can't see the secondary storage and thus I can't download templates/ISOs. The dashboard says 0kb of 0kb in use for secondary storage.

I have tried having the secondary storage as local to the cloudstack management instance (whilst setting the use.local global setting to true). I have also tried setting up a new host and setting that up as the NFS share and it did not work.

I have checked in both instances that the shares I have made are mountable - and they are. I have also seeded them with the template VM by running the command outlined in the installation guide. Both places I set to be secondary storage had ample space available - 1 greater than 200GB. The other around 70GB. I have also restarted the management machine a few times.

Any help would be much appreciated!

bean
  • 1,091
  • 1
  • 12
  • 23
  • I understand the request to close but there are more questions about cloudstack on SO than on serverfault so I thought it was the correct place. Is there a better stackexchange to ask this question on? – bean Aug 15 '14 at 10:30

1 Answers1

2

You need secondary storage enabled in order to supply templates to your hosts. The simplest way to achieve that is to create an NFS export that is available to the host. I usually do it in the host it self. In your case that would be the XenServer. Then in the management server add the secondary storage in: Infrastructure -> Secondary Storage -> Add Secondary Storage.

Secondary storage is provided by a dedicated system VM. Once you add a secondary storage, CloudStack will create a system VM for that. Start by checking the status of the system VMs in: Infrastructure -> System VMs

The one you are looking for should be called Secondary Storage VM.

It should be running and the agent should be ready (two green circles). If the agent is not ready, first ssh to your XenServer host and then to the system VM using the link local IP (you can see the IP in the details of the VM) with the following command:

ssh -i /root/.ssh/id_rsa.cloud -p 3922 LIKN_LOCAL_IP_ADDRESS

Then in the system VM, run a diagnostic tool to check what could be wrong:

/usr/local/cloud/systemvm/ssvm-check.sh
Miguel Ferreira
  • 1,282
  • 11
  • 27
  • Thank you for the insight. I hadn't appreciated this! I thought it was just an NFS share. I'd wondered why it needed seeding with that template! I will check this now and see where things stand. – bean Aug 15 '14 at 12:27
  • I don't seem to have that system template running. Here is all I have (Console proxy vm) http://i.imgur.com/ttm3Kje.png . If I go to start up a new instance from templates I can't see the system template (despite it being in the template list). I'm guessing this is because I have no secondary storage! – bean Aug 15 '14 at 12:33
  • No, it's not in a ready state. Agent state is just disconnected. – bean Aug 15 '14 at 12:38
  • I would say you have a netwoking problem. The console proxy VM is also a system VM and should have the `ssvm-check.sh` diagnosis tool. Run that and it should point you to what's wrong. I will update the answer regarding not having a secondary storage. – Miguel Ferreira Aug 15 '14 at 12:56
  • Ahhh. Thank you for that edit. From the installation guide I misunderstood and thought my secondary storage should be on the CloudStack mgmt server. I think I may just reinstall all and try again! Thanks again. – bean Aug 15 '14 at 13:04
  • I am having similar problem. System VM count is showing zero in cloudstack management UI. Why could be that? – Muhammad Adeel Zahid Apr 06 '17 at 15:20
  • It could be that the SystemVM is failing to start. CloudStack should periodically try to start 2 SystemVMs (for sec storage and console proxy), and that is usually visible on the UI. First you see them spinning up, and then they go into an error state and are removed. In that case you need to figure out why are they failing to start. – Miguel Ferreira Apr 14 '17 at 16:47