Search code examples
openstackdevstack

Error in re-running stack.sh


I am new to openstack. I have recently done a single node openstack installation in my ubuntu virtual machine from the following tutorial:-

http://www.openflowhub.org/display/floodlightcontroller/Install+Floodlight+and+OpenStack+on+Your+Own+Ubuntu+VM

And i was able to access the services through Horizon. But after a restart, i was not able to access anything of openstack. So i had to re-run stack.sh during which i got the following error in the quantum net-create command :-

Usage: quantum [OPTIONS] <command> [args]
quantum: error: no such option: --tenant_id

Can anyone please help me with this?


Solution

  • When working with devstack, it's important to understand that it is not intended to be a fully functioning openstack installation. It is a tool for developers. As such, the devstack installer script "stack.sh" functions on the assumption that the devstack deployment is ephemeral.

    Basically there is no persistence. When you reboot, the software will still reside locally on the disk, but there are no supporting startup routines setup in init or upstart or what have you.

    The accepted method of dealing with this is to run "unstack.sh" which will remove the previous devstack configuration. Then running "stack.sh" again after that.

    As an addendum.

    It has been my experience that updating devstack BEFORE running an unstack.sh on a devstack environment can corrupt that environment. Basically if the cleanup logic adapts to a new way of installation it won't be suited to unstack a previous versions stack.sh.

    So be careful. My preference is running devstack in a kvm instance. I keep a backup qcow with a known working config available to swap over to if i corrupt something. Again this is a dev environment. You aren't supposed to fix it when it breaks. You are supposed to nuke it from orbit and start over. This is the cloud way.