Re: [openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-08-02 Thread Adam Heczko
Agree that SERVICE_TOKEN usage eradication will be probably long standing process, but IMO radosgw should follow usual way of managing Openstack service interactions. Usually when service wants to integrate with OpenStack, an appropriate user with role "admin" is created. I believe that for radosgw

Re: [openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-07-30 Thread Oleksiy Molchanov
Update from Radoslaw Zarzynski --- Hi, I'm afraid that eradication of OS_SERVICE_TOKEN won't be quick nor painless process due to dependencies. We would need to identify and fix all applications that requires this auth method. For example, Ceph RADOS Gateway (radosgw) currently requires [1]

Re: [openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-07-29 Thread Konstantin Danilov
Would send ceph estimation tomorrow. Yet estimation != ETTA On Wed, Jul 29, 2015 at 12:27 AM, Sergii Golovatiuk wrote: > Hi, > > Let's ask our Ceph developers how much time/resources they need to implement > such functionality. > > -- > Best regards, > Sergii Golovatiuk, > Skype #golserge > IRC #

Re: [openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-07-28 Thread Sergii Golovatiuk
Hi, Let's ask our Ceph developers how much time/resources they need to implement such functionality. -- Best regards, Sergii Golovatiuk, Skype #golserge IRC #holser On Tue, Jul 28, 2015 at 11:21 PM, Andrew Woodward wrote: > It's literally how radosgw goes about verifying users, it has no schem

Re: [openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-07-28 Thread Andrew Woodward
It's literally how radosgw goes about verifying users, it has no scheme of using a user or working with auth-tokens. It would have to fixed in the ceph-radosgw codebase. PKI tokens (which we don't use) rely on this less, but its still used. On Tue, Jul 28, 2015 at 2:16 PM Sergii Golovatiuk wrote:

Re: [openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-07-28 Thread Sergii Golovatiuk
Why can't radosgw use own own credentials? If it's technical debt we need to put it on plate to address in next release. -- Best regards, Sergii Golovatiuk, Skype #golserge IRC #holser On Tue, Jul 28, 2015 at 10:21 PM, Andrew Woodward wrote: > Keystone authtoken is also used by radosgw to vali

Re: [openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-07-28 Thread Andrew Woodward
Keystone authtoken is also used by radosgw to validate users On Tue, Jul 28, 2015 at 10:31 AM Andrew Woodward wrote: > IIRC the puppet modules, and even the heat domain create script make use > of the token straight from the config file. It not being present could > cause problems for some of th

Re: [openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-07-28 Thread Andrew Woodward
IIRC the puppet modules, and even the heat domain create script make use of the token straight from the config file. It not being present could cause problems for some of the manifests. We would need to ensure that their usage is minimized or removed. On Tue, Jul 28, 2015 at 9:29 AM Sergii Golovat

Re: [openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-07-28 Thread Sergii Golovatiuk
Hi Oleksiy, Good catch. Also OSTF should get endpoints from hiera as some plugins may override the initial deployment settings. There may be cases when keystone is detached by plugin. -- Best regards, Sergii Golovatiuk, Skype #golserge IRC #holser On Tue, Jul 28, 2015 at 5:26 PM, Oleksiy Molchan

[openstack-dev] [fuel] OS_SERVICE_TOKEN usage in Fuel

2015-07-28 Thread Oleksiy Molchanov
Hello all, We need to discuss removal of OS_SERVICE_TOKEN usage in Fuel after deployment. This came from https://bugs.launchpad.net/fuel/+bug/1430619. I guess not all of us have an access to this bug, so to be short: # A "shared secret" that can be used to bootstrap Keystone. # This "token" does