Description of problem: UPI AWS installs renders manifests for machines. In UPI modes, usage of the machine API is a choice, and I do not think we can render proper manifests without collecting more data. How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: machine-api resources are rendered in an install Expected results: do not render machine api resources, or make dynamic compute an option in an install-config. Additional info:
(In reply to Derek Carr from comment #0) > Description of problem: > > UPI AWS installs renders manifests for machines. In UPI modes, usage of the > machine API is a choice, and I do not think we can render proper manifests > without collecting more data. > > How reproducible: > > Steps to Reproduce: > 1. > 2. > 3. > > Actual results: > machine-api resources are rendered in an install > > Expected results: > do not render machine api resources, or make dynamic compute an option in an > install-config. install-config currently supports 0 compute: ``` apiVersion: v1beta3 baseDomain: tt.testing compute: - name: worker replicas: 0 controlPlane: name: master replicas: 1 ``` > Additional info: And users are free to remove the machine objects for control plane if they choose. ``` create manifests rm -rf openshift/99_openshift-cluster-api_master-machines-*.yaml create ignition-configs ```
*** Bug 1700255 has been marked as a duplicate of this bug. ***
This is not a BetaBlocker, for now we'll document the need to step through manifest generation and removal of machine manifests from AWS UPI installs as described in Comment 1.
Scott, will you either dupe or reassign the bug to doc?
(In reply to Kathryn Alexander from comment #4) > Scott, will you either dupe or reassign the bug to doc? Yes, there's some due diligence we need to do on our side and we'll pass this over once we're done.
Ok, installer#1649 tested out for me locally, so I'm moving this to the Documenation component.
I added two commits to https://github.com/openshift/openshift-docs/pull/14241 to add steps to remove the control plane manifest and update the CloudFormation templates. At this point, I think that the cluster-controlled compute nodes should be handled separately.
Follow the PR with r4.1.0-rc.0 payload, have a try, still reproduced. The 1st issue: Should be `rm -f openshift/99_openshift-cluster-api_master-*.yaml`, but not `rm -f openshift/99_openshift-cluster-api_master-machines-*.yaml` or else would hit such error: $ ./openshift-install create ignition-configs --dir ./upi_2019-04-24-23-57-58 FATAL failed to fetch Bootstrap Ignition Config: failed to load asset "Master Machines": master machine manifests are required if you also provide openshift/99_openshift-cluster-api_master-user-data-secret.yaml The 2nd issue after workaround the 1st issue: + ./openshift-install create manifests --dir ./upi_2019-04-25-02-31-11 WARNING There are no compute nodes specified. The cluster will not fully initialize without compute nodes. INFO Consuming "Install Config" from target directory + echo 'Remove the control-plane Machines from the manifest assets.......' Remove the control-plane Machines from the manifest assets....... + rm -f ./upi_2019-04-25-02-31-11/openshift/99_openshift-cluster-api_master-machines-0.yaml ./upi_2019-04-25-02-31-11/openshift/99_openshift-cluster-api_master-machines-1.yaml ./upi_2019-04-25-02-31-11/openshift/99_openshift-cluster-api_master-machines-2.yaml ./upi_2019-04-25-02-31-11/openshift/99_openshift-cluster-api_master-user-data-secret.yaml + echo 'Generate ignition files.......' Generate ignition files....... + ./openshift-install create ignition-configs --dir ./upi_2019-04-25-02-31-11 INFO Consuming "Common Manifests" from target directory INFO Consuming "Openshift Manifests" from target directory INFO Consuming "Worker Machines" from target directory Follow upi-on-aws install doc, master machines are still listed: # oc get machine -n openshift-machine-api NAME INSTANCE STATE TYPE REGION ZONE AGE jialiu-upi2-rlv4l-master-0 m4.xlarge us-east-2 us-east-2a 27m jialiu-upi2-rlv4l-master-1 m4.xlarge us-east-2 us-east-2b 27m jialiu-upi2-rlv4l-master-2 m4.xlarge us-east-2 us-east-2c 27m
Trevor, will you PTAL at issue 2?
(In reply to Johnny Liu from comment #9) > The 1st issue: > Should be `rm -f openshift/99_openshift-cluster-api_master-*.yaml`, but not > `rm -f openshift/99_openshift-cluster-api_master-machines-*.yaml` or else > would hit such error: > $ ./openshift-install create ignition-configs --dir ./upi_2019-04-24-23-57-58 > FATAL failed to fetch Bootstrap Ignition Config: failed to load asset > "Master Machines": master machine manifests are required if you also provide > openshift/99_openshift-cluster-api_master-user-data-secret.yaml I cannot reproduce, and have posted some notes on this to [1]. Seeing if I can reproduce your second point now... [1]: https://github.com/openshift/installer/pull/1649#discussion_r278710662
I removed my previous installer binary, and extract it. The same version. While tThis time, I can not reproduced either. Really interesting. $ oc adm release extract --command='openshift-install' registry.svc.ci.openshift.org/ocp/release:4.1.0-rc.0 $ ./openshift-install version ./openshift-install v4.1.0-201904211700-dirty built from commit f3b726cc151f5a3d66bc7e23e81b3013f1347a7e release image registry.svc.ci.openshift.org/ocp/release:4.1.0-rc.0 $ rm -f ./test1/openshift/99_openshift-cluster-api_master-machines-*.yaml $ ./openshift-install create ignition-configs --dir test1 INFO Consuming "Master Machines" from target directory INFO Consuming "Openshift Manifests" from target directory INFO Consuming "Worker Machines" from target directory INFO Consuming "Common Manifests" from target directory Follow upi-on-aws install doc, master machines are not listed any more. # oc get machine -n openshift-machine-api No resources found.
The AWS UPI docs are live.