[Ci-users] Reclaiming old Openstack-Puppet nodes in ci.centos.org

Wed Oct 14 15:15:50 UTC 2015
Karanbir Singh <kbsingh at centos.org>

Hash: SHA1


On 14/10/15 16:08, Gabriele Cerami wrote:
> On 14 Oct, Brian Stinson wrote:
>> Hi All,
>> Looks like we have a few hosts assigned to the Openstack-Puppet
>> project that have been hanging around for a while. I'd like to
>> add this project to our automatic reaper script.
> Hi,
> You can return all of them at any time, but before you do that is
> there a way to see a list of hosts assigned to a user ? I'd like
> to know how many host we left pending, and in what times, if I have
> to correct the way we're returning them, and not leaving anything
> behing. In any case, how is the reaper script working ? checks for
> ssh logins time ? how much time do we have until hosts are
> forcefully returned ?

At the moment, its very casual - you get atleast 24 hrs + time of day
in date you requested the machine. so if you ask for a node at 00:00
then you get 48 hrs before the machine is recovered.

If you request a machine at 23:59 on a day then you get 24 hrs.

The clock starts from the time of request for node, and the entire
session is handled as a single block ( ie. if you ask for 3 nodes,
they will all be recovered at the same time ).

I will modify the /Inventory call to allow folks to specify their own
api key, and get a host list. Let me confirm once that is done.

In most cases, the hosts get left behind due to a crash or a unclean
shutdown in the jenkis job itself.

At some point in the future, once we start getting closer to capacity
limits, we want to try and bring the deployed time down to 6 to 8 hrs
per host and recover it that way ( but we are a way from hitting those
yet ).


- -- 
Karanbir Singh, Project Lead, The CentOS Project
+44-207-0999389 | http://www.centos.org/ | twitter.com/CentOS
GnuPG Key : http://www.karan.org/publickey.asc
Version: GnuPG v2.0.22 (GNU/Linux)