[FCOS] Control amount of replicas in etcd-quorum-guard deployment#1708

Merged

openshift-merge-robot merged 1 commit intoopenshift:fcosfrom

vrutkovs:fcos-quorum-guard-scale

May 20, 2020

Contributor

vrutkovs commented May 1, 2020 •

edited

Loading

- What I did

Moved etcd-quorum-guard deployment under MCO control (instead of CVO).
Updated the code which applies MCD daemonset to apply etcd-quorum-guard deployment
Updated amount of replicas in etcd-quorum-guard deployment - if CEO's unsupported non-HA option is enabled MCO would create a deployment with one replica.
etcd-quorum-guard now uses pod image instead of cli

This allows provisioning single node clusters.

- How to verify it
Create a cluster with 0 workers, 1 master and apply CEO's manifest to allow less than 3 workers(openshift/cluster-etcd-operator#279).

TODO:

Check if cliImage can be avoided and infraImage can be reused instead

This is submitted for fcos branch only, as master's implementation would depend on scalable control plane enhancement

/cc @LorbusChris

Fixes okd-project/okd#162

openshift-ci-robot requested a review from LorbusChris

May 1, 2020 12:06

openshift-ci-robot added the approved label

vrutkovs force-pushed the fcos-quorum-guard-scale branch 2 times, most recently from 92f0a31 to 90a7f31 Compare

May 1, 2020 13:17

vrutkovs mentioned this pull request

etcd-quorum-guard should not attempt to deploy 3 replicas if useUnsupportedUnsafeNonHANonProductionUnstableEtcd is set #1695

Closed

vrutkovs force-pushed the fcos-quorum-guard-scale branch 2 times, most recently from aef70ca to 6930a61 Compare

May 2, 2020 05:51

Contributor

LorbusChris commented May 5, 2020 •

edited

Loading

I'd rather not introduce this now that are so close to merging fcos with master and instead wait for implementation of openshift/enhancements#292

Contributor Author

vrutkovs commented May 5, 2020

That's for 4.4, the enhancement won't be backported there anyway - and we can branch fcos-4.4 later

Member

ashcrow commented May 5, 2020

/cc'ing @cgwalters @dustymabe as well for review

Contributor

LorbusChris commented May 5, 2020

this is only needed for OKD CRC, right?

Contributor Author

vrutkovs commented May 5, 2020

That's not required for CRC, but with this PR you can setup a single-node cluster (still having bootstrap node), but that would be a full blown cluster, no operators disabled / ripped. This cluster, of course, cannot be updated

Contributor

LorbusChris commented May 7, 2020

Just noting that this will need to be handled differently in master from 4.6 on, as we won't have an fcos fork anymore by then. Otherwise we'll have to regress on this in OKD 4.6.
/lgtm

openshift-ci-robot assigned LorbusChris

openshift-ci-robot added the lgtm label

Contributor Author

vrutkovs commented May 7, 2020

/refresh

1 similar comment

Contributor Author

vrutkovs commented May 11, 2020

/refresh

Contributor

openshift-bot commented May 11, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

2 similar comments

Contributor

openshift-bot commented May 11, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 11, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor Author

vrutkovs commented May 11, 2020

/hold

level=fatal msg="failed to initialize the cluster: Cluster operator machine-config is reporting a failure: Failed to resync 0.0.1-2020-05-11-102411 because: failed to execute template: template: manifests/etcdquorumguard_deployment.yaml:54:24: executing \"manifests/etcdquorumguard_deployment.yaml\" at <.Images.KubeClientAgent>: can't evaluate field KubeClientAgent in type *operator.RenderConfigImages"

openshift-ci-robot added the do-not-merge/hold label

vrutkovs force-pushed the fcos-quorum-guard-scale branch from 6930a61 to 938995e Compare

May 11, 2020 11:42

openshift-ci-robot removed the lgtm label

vrutkovs force-pushed the fcos-quorum-guard-scale branch 6 times, most recently from e306b91 to 197c397 Compare

May 11, 2020 15:24

openshift-ci-robot added the lgtm label

Contributor

openshift-bot commented May 18, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

13 similar comments

Contributor

openshift-bot commented May 18, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 18, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 18, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 18, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 18, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 19, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 19, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 19, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 19, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 19, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 19, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 19, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor

openshift-bot commented May 19, 2020

/retest

Please review the full test history for this PR and help us cut down flakes.

Contributor Author

vrutkovs commented May 19, 2020

/hold

PDB is not valid :(

openshift-ci-robot added the do-not-merge/hold label


          Control amount of replicas in etcd-quorum-guard

884f76b

MCO should apply etcd-quorum-guard deployment instead of CVO.
It also controls the number of replicas in this deployment: it would
scale 1 replica if CEO's useUnsupportedUnsafeNonHANonProductionUnstableEtcd
option is enabled.

This allows creating single node clusters

vrutkovs force-pushed the fcos-quorum-guard-scale branch from 5c04c20 to 884f76b Compare

May 19, 2020 08:08

openshift-ci-robot removed the lgtm label

Contributor Author

vrutkovs commented May 19, 2020

/retest

Storage flakes

Contributor Author

vrutkovs commented May 19, 2020

/test e2e-aws

1 similar comment

Contributor Author

vrutkovs commented May 20, 2020

/test e2e-aws

Contributor Author

vrutkovs commented May 20, 2020

/hold

Contributor Author

vrutkovs commented May 20, 2020

/hold cancel

@LorbusChris ready for another review (added a default setting for etcd-quorum-guard replicas)

openshift-ci-robot removed the do-not-merge/hold label

Contributor

LorbusChris commented May 20, 2020

/lgtm

openshift-ci-robot added the lgtm label

Contributor

openshift-ci-robot commented May 20, 2020

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: LorbusChris, vrutkovs

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

~~OWNERS~~ [LorbusChris,vrutkovs]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

openshift-merge-robot merged commit 0e34bdf into openshift:fcos

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels