-
Notifications
You must be signed in to change notification settings - Fork 126
General: FAQ
This section groups all common problems you may have installing or using Archipel.
First of all, before continuing, update your copy or Archipel Agent with # easy_install -U archipel-agent Then try using http://app.archipelproject.org for the GUI or get the latest nightly build from http://nightlies.archipelproject.org, and try again. If problem is still present, you can continue to read this FAQ.
You can use:
- Chromium (All platforms): best choice
- WebKit (Mac OS): the other best choice
- Chrome (All platforms)
- Safari (Mac OS)
- Firefox (All platforms)
- ReKonq
Are you kidding?
501 error means "Not Implemented". This is because your agent can't process the request (i.e. your request has not been caught by any XMPP handlers). There are several reasons:
- One of the agent's modules has not been loaded for some reason. Carefully read the beginning of archipel.log to track down
ERRORmessages in the early initialization process. Most of these errors are due to missing configuration tokens. - Otherwise, you are not up to date. Even if you think you are, you are not. Try to Reset EGG Installation.
Install the xmpppy package with easy_install, the package in Ubuntu is too old.
When I want to launch Archipel GUI and I look at the Debug Console, I have a lot of error telling "Info.plist" not found
This is not a problem. This is the way Cappuccino determines if a folder is a bundle or a framework or just a standard folder. It tries to get Info.plist from folders and this can result as a 404 error displayed in the console.
Your XMPP admin account SHOULD to be called "admin" (for example admin@myserver.com). Otherwise, you need to add your admin account name in to the Resources/admin-accounts.js file, in the client folder being served by your web server.
You get something like ERROR ::2011-01-14 16:09:09::<archipelHypervisor.TNArchipelHypervisor instance at 0x14ab5f0>.iq_get_permission: exception raised is : No connection has been defined for this thread or process. This mean you are using an outdated version of python SQLAlchemy. You MUST use the easy_install package of SQLAlchemy, not the one eventually packaged in your distribution. Remove any installation of this egg, and reinstall it with
# easy_install SQLAlchemy
I have successfully installed archipel, I'm connected, and I have added the hypervisor in the roster, but it is displayed as "Offline"
You probably have manually created the hypervisor account using ejabberdctl which is bad. You have now two solutions:
- change the account of the hypervisor of the token
hypervisor_xmpp_jidin thearchipel.conffile and restart Archipel. - unregister the account using
ejabberdctl unregister hypervisor FQDNand restart Archipel.
YOU MUST NOT CREATE AN XMPP ACCOUNT FOR ARCHIPEL ENTITIES. Accounts can be manually created for users.
export PYTHONDONTWRITEBYTECODE as empty, and rerun easy_install # export PYTHONDONTWRITEBYTECODE= # easy_install archipel-agent
The New VM button is a future feature. Do not use it at the moment. To create a VM, go to a hypervisor's, "Virtual Machine" tab and click on the "+" button.
For hackerz: You can use our fork of ejabberd with needed fixes, see https://github.com/primalmotion/ejabberd/tree/archipel-v2.1.8
Most of the time, you will get some error in the console saying "Unable to connect to 127.0.0.1:69XX". This is because for some reason, the Archipel Agent hasn't be able to find its public IP. To fix this, edit /etc/archipel/archipel.conf, and change the value of the token machine_ip = auto to machine_ip = X.X.X.X where X.X.X.X is your hypervisor's public IP. Then restart the agent and then the virtual machines from Archipel GUI. It should work.
Also note that the VNC connections are initiated from your browser using a websocket proxy. This proxy uses a port equal to VNC-port + 1000. So if your VM uses the port 5901, Archipel will connect to the proxy listening to 6901. Obviously, you must be able to connect to these ports from the computer that runs the browser that runs Archipel GUI.
If you are using centos 6.0 (http://lists.centos.org/pipermail/centos-virt/2011-October/002649.html), taking a snapshot through the binary savevm isn't supported with version 0.8.1 of libvirt. You can either wait for the release of centos 6.1, or use the Continuous Realease repo and benefit from libvirt 0.8.7. Have a look here: http://lists.centos.org/pipermail/centos-announce/2011-September/018078.html
Follow these 5 steps:
- yum install centos-release-cr
- add "priority 1" at the end of /etc/yum.repos.d/CentOS-CR.repo
- yum install libvirt
- check version of libvirt: rpm -qva | grep libvirt
- service archipel restart
It should work now.
You need to change the cache mode of your drives to "None" in the virtual machine definition. This is mostly due to a bug in some versions of libvirt.
There are some errors while talking to the libvirt daemon: look in /var/log/archipel-startup.log for libvirt errors.
Verify you have installed qemu-kvm for example...
Restart Archipel agent on the hypervisor. This bug happens sometimes after calling archipel-xxxnode --create commands.
Two choices, either you screwed the vCard of the xmpp entity (check archipel.conf vCard section), or the version of ejabberd you are using has issue with vCards.
To check the content of the vCard you can do from the ejabberd server
ejabberdctl get_vcard $USER `hostname` ROLEReplace $USER with a proper user you find using ejabberdctl registered_users.
Valid roles are hypervisor, virtualmachine, user or central-agent.