We talked about a number of things at our flock meeting, and a few
things we didn't get to.
Here's a summary. Please chime in if you think I summarized wrong or
you would like to add things or revist something. ;)
* Jenkins
- Finish migrating to the new cloud instance with the packaged jenkins
- Will will weaken the wording on the wiki page. We still don't want
people to depend on our jenkins for release critical stuff, but we
will want it to be more supported.
- Add nagios checks so we know when it's down/needing fixing.
* Finishing the puppet migration
- This week I will create the lockbox01 replacement (batcave01).
- Once we have that all ansiblized and ready, we will change it's
ip's to the current lockbox01 ips.
- Any machines still in puppet at that point will be marked "Don't
touch" and we will try and get them migrated. This will likely be
collab03 (mailing lists migrating off it to mailman01), and
hosted03 (until we finish rhel7 packaging work and migrate it).
- gobby (or something better if we can find it) will migrate to some
other host.
* bodhi2
- Deploy tomorrow.
- Once things are working we can retire releng04 and repele01
* Openstack plans:
- Move all instances to new cloud
- Add old cloud nodes to new cloud
- Install two old nodes with openstack Icehouse
- Upgrade the test instance to openstack Kilo(?)
- Upgrade main cluster to Kilo
* Continuous deployment
- May not have enough advantages to actually do continuous
deployment but the steps that enable it are useful
- work on those steps and then revisit in 6 months
- Automate ansible creation of new hosts
- Automate deployment of our apps
- Upgrade playbooks (Take service out
- Automate sign and import of packages
* Apprentice / on boarding discussion -
* make a set of quests to point people to:
* update CSI info
* Update docs templates
* others?
* Review in meetings easyfixes.
* Try and do a 'technical debt workdown' week after F23 is released.
Possibly also try 'docs update' or the like.
- Try one every 3 months or so.
* Docker/container prototype
* John Stanley containerized a sticky notes instance in the cloud for
us to look at.
Things we wanted to talk about, but did not get to:
* Released app process and repos. We actually did talk about this, but
didn't actually make any decisions. It sounded like we wanted apps to
have all their dependencies in EPEL, but they could be in the
infrastructure repo if they were fast moving?
* FAS3. Whats the timeline for migration?
* single-sign-on/single-sign out.
* Auth protocol
* Hubs development / prototyping
Anything else?
kevin
Attachment:
pgp9IQrfAoERW.pgp
Description: OpenPGP digital signature
