Open Stack

On Wed, 2019-07-10 at 07:50 +0100, Chris Dent wrote:
> On Wed, 10 Jul 2019, Guannan GN2 Sun wrote:
> 
> > I meet some problem when deploying devstack(from
> > https://github.com/openstack/devstack master branch) on Ubuntu
> > 16.04. It seems something is wrong with placement-api error
> > message as following:
> > 
> > 
> > curl -g -k --noproxy '*' -s -o /dev/null -w '%{http_code}%' http://10.240.24.138/placement
> > 
> > [[503 == 503]]
> > 
> > [ERROR] /opt/stack/devstack/lib/placement:156 placement-api did not start
> > 
> > 
> > However when I check its status using "systemctl status
> > devstack at placement-api", it is active and running. I also change
> > to "stein" branch and try to deploy again, but still meet the same
> > problem. Does someone meet similar issue before or could someone
> > help me to debug this issue? Below is my local.conf file. Thank
> > you!
> 
what cdent said below is also true but just wanted to highlight
that master of devstack is not intendeted to be run with ubuntu 16.04
we move to 18.04 some thime ago upstream so if there have been any deviations
between 16.04 and 18.04 dont expect master of devestack to support both.

> Look in /etc/apache2/sites-enabled/placement-api.conf (and any other
> files in the same directory) and make sure you only have one proxy
> configuration for the connection between apache2 and the uwsgi
> process that is running under systemd.
> 
> What's could be happening is that though you have placement running,
> apache is trying to talk to the wrong thing.
> 
> You can either:
> 
> * clean up the placement-api.conf file so that it only has one entry
>    and restart apache2
> 
> * unstack.sh, remove the files in /etc/apache2/sites-enabled and
>    /etc/apaceh2/sites-available, and rerun stack.sh
> 
> This happens when there are repeated runs of stack.sh on the same
> host with insufficient cleanup between. This probably means there's
> a bit of a bug in lib/placement that we could fix so that it cleans
> up after itself better.
> 
> I hope this is helpful. If this wasn't it and you figure out what
> was causing it please post the solution.
>