Reinstalling the IPA server in staging made the enrollments of IPA
clients against the old instance stale. This playbook detects this,
cleans things up and enrolls hosts again.
Signed-off-by: Nils Philippsen <nils@redhat.com>
Pull in redis/docker-entrypoint.sh: listen on all interfaces
b239f79e58
This should make redis accessible on the redis service and not just
locally in the redis container.
When we setup things we didn't setup fedmsg in iad2 staging.
Now we are using it, so we should configure it until we get rid of it.
Signed-off-by: Kevin Fenzi <kevin@scrye.com>
I think the only reason this worked in phx2 was because we also had a
firewall rule that was blocking the staging subnet on all prod machines.
Instead of doing that here, lets just make fedmsg only deal with the
staging hosts in staging. The only thing this might break is things that
aren't in staging_friendly that should be.
Signed-off-by: Kevin Fenzi <kevin@scrye.com>
Match the environment variable in the deploymentconfig to that used
in the configmap (REDIS_PASSWORD)
Signed-off-by: Owen W. Taylor <otaylor@fishsoup.net>
The 'containerNames' parameter was incorrectly cut-and-pasted from
the flatpak-indexer-differ deployment config.
Signed-off-by: Owen W. Taylor <otaylor@fishsoup.net>
Should be ImageStreamTag not ImageStream. We ended up with a destination
of flatpak-indexer:latest:latest, when a default tag of :latest was added.
Signed-off-by: Owen W. Taylor <otaylor@fishsoup.net>
This should set up the bridge and controls on the new vmhosts in
RDU-CC so that we can deploy a dhcp host which will allow mgmt to be
brought up for power and similar hardware.
Signed-off-by: Stephen Smoogen <ssmoogen@redhat.com>
* Output the build to the correct imagestream - it was outputing
to the flatpak-indexer-tardiff stream from a cut-and-paste mistake
* Add a ImageChange trigger for the from: image - we want to trigger
a rebuild if the ubi8/python-38 image changes by itself, even if
the flatpak-indexer-tardiff image where we get the tardiff binary from
isn't rebuilt.
Signed-off-by: Owen W. Taylor <otaylor@fishsoup.net>
* Specify a volumeName for the persistant volume claims (on initial deploy
to staging, these PVC's got bound to random volume...)
* Name the PVC's to exactly match the volume names, including a -stg suffix
for staging
* Reduce the requested size for the redis volume to 1Gi.
The deployment config for flatpak-indexer referenced a secret volume that
in the Red Hat internal deployment of flatpak-indexer holds a client certificate,
but is not needed for Fedora.
Signed-off-by: Owen W. Taylor <otaylor@fishsoup.net>
