(feat): Extra Valkey config by utdrmac · Pull Request #44 · valkey-io/valkey-operator

utdrmac · 2026-01-09T23:12:05Z

This feature request implements the ability for users to supply their own configuration for Valkey as an inline stanza when deploying a new cluster. The user configuration is appended to the required base configuration, which is created as a configMap. Since the operator manages the configMap, any changes to the config are reconciled. Additionally, users cannot override the required cluster settings.

This PR implements the ability for users to supply additional configuration for Valkey. Any provided parameters are appended to default configuration required by the operator Signed-off-by: utdrmac <matthew.boehm@percona.com>

Signed-off-by: utdrmac <matthew.boehm@percona.com>

sandeepkunusoth

LGTM

sandeepkunusoth

lint errors and e2e errors

Signed-off-by: utdrmac <matthew.boehm@percona.com>

utdrmac · 2026-01-10T03:12:22Z

ugh. why didn't lint/fmt/vet catch those?

Signed-off-by: utdrmac <matthew.boehm@percona.com>

jdheyburn

Minor comments, keen to hear opinions, otherwise LGTM

Signed-off-by: utdrmac <matthew.boehm@percona.com>

utdrmac · 2026-01-12T21:48:26Z

e2e still fails due to #43

sandeepkunusoth · 2026-01-12T22:04:35Z

e2e still fails due to #43

no this is different issue than 43. its failing because of events limitation. maybe as u have added new emit event other events are not coming. https://github.com/valkey-io/valkey-operator/actions/runs/20936033780/job/60159472854?pr=44.

check this PR description #37. maybe you can update events check accordingly. currently we are not asserting for all events due to events rate limit.

utdrmac · 2026-01-12T22:29:42Z

I only added 1 new event which isn't triggered during the e2e. When I run e2e on my local kind cluster, I don't get that failure; I get the same failure as #43.

sandeepkunusoth · 2026-01-12T22:46:00Z

I only added 1 new event which isn't triggered during the e2e. When I run e2e on my local kind cluster, I don't get that failure; I get the same failure as #43.

no i don't think its related. you can check the issue description failure test case and logs are different for both of these errors. i have neven seen failure of valkey creation CR test case in main branch. if in case we are seeing, we need to create different issue.

this PR is failing this test "Manager when a ValkeyCluster CR is applied [It] creates a Valkey Cluster deployment"

43 issue is with other test "Manager when a ValkeyCluster experiences degraded state [It] should detect and recover when a deployment is deleted" https://github.com/valkey-io/valkey-operator/actions/runs/20806446577/job/59761668293

quick fix: we may need to remove assertions of some of the events.

valkey-operator/test/e2e/e2e_test.go

Line 405 in 44d674b

    
           g.Expect(output).To(ContainSubstring("ClusterMeet"), "ClusterMeet event should appear in describe")

bjosv

LGTM

bjosv · 2026-01-15T12:01:13Z

Rebasing (or add merge commit of main) should fix the e2e test now.

Signed-off-by: utdrmac <matthew.boehm@percona.com>

jdheyburn · 2026-04-02T05:44:24Z

+			Tolerations:        cluster.Spec.Tolerations,
+			Exporter:           cluster.Spec.Exporter,
+			Containers:         cluster.Spec.Containers,
+			UsersConfigMapName: getConfigMapName(cluster.Name),


I wonder if UsersConfigMapName implies it is related to Users, in a similar way that UsersACLSecretName does. Or does Users mean the human deploying the operator?

Users, as the human deploying the operator. There are some config params that the operator sets, and others that the users would set.

My interpretation was that Users here meant for ACLs. If it is ambiguous maybe it should be removed?

jdheyburn · 2026-04-02T05:57:32Z

+	// This hash should be updated whenever the contents of either script changes, which would
+	// coincide with operator version bump.
+	// $ cat internal/controller/scripts/{liveness-check.sh,readiness-check.sh} | sha256sum
+	scriptsHash = "8531132f52ac311772dfcb45c107c34ab05e719a0df644cc332512277b564346"


Is there not a way we can have this calculated for us?

The goal was to not have the operator (re)calculating sha256 on static file contents every 30s. As for an "outside" solution, I could probably do something similar to how version numbers are done, by linking the value to the variable during compile time?

How do other operators do this? Could we perform a faster check before the sha256?

has_changed = false if len(current) != len(desired): has_changed = true elif sha256(current) != sha256(desired): has_changed = true

I can speak for 3 of Percona's operators (mysql, pxc, mongo): The liveness, readiness, and health check scripts are part of the operator docker image. The image is used as init-container, where a separate entrypoint runs to copy the scripts into the running pod. The init container finishes, then the main container, using the same image with default entrypoint, launches as a regular container and the scripts are there to be used.
This has some benefits: 1) no checksum calculations, 2) nothing to 'embed', and 3) upgrades to the script become automatic new release/version of the operator.
Once we decide on if we will be making an operator-specific image (ie: for including modules), perhaps we could use a similar model/process?

That could work, but until we decide on that - could we instead calculate the sha256sum on operator boot? Given that the scripts would be coupled with the operator version, we would then be able to avoid calculating the sha on every reconcile. How does that sound?

Sounds doable!

jdheyburn · 2026-04-02T05:59:19Z

+var scripts embed.FS
+
+func getConfigMapName(clusterName string) string {
+	return clusterName + "-config"


Can we prepend valkey- to this? I'm thinking in case there is a conflict and we can minimise that.

Yup. I'll push that shortly.

jdheyburn · 2026-04-02T06:01:18Z

+
+// Create or update a default valkey.conf
+// If additional config is provided, append to the default map
+func (r *ValkeyClusterReconciler) upsertConfigMap(ctx context.Context, cluster *valkeyiov1alpha1.ValkeyCluster) error {


I've started refactoring other upserts to use controllerutil.CreateOrUpdate, could the same be done here for consistency? It might help reduce some lines too.

Probably. I will look into it.

I wonder if this would help with the script-hash that is being set in the PR today.

jdheyburn · 2026-04-02T06:22:55Z

+
+	// Build the config
+	var configBuilder strings.Builder
+	configBuilder.Grow(len(specConfig) * averageParameterLength)


I wonder if the use of Grow here is prematurely optimising, would it simplify the logic if we removed it?

It would simplify the logic by 1 line. I read that this pre-allocation was a simple memory optimization if the size of the eventual string was somewhat known.

Signed-off-by: utdrmac <matthew.boehm@percona.com>

utdrmac · 2026-04-09T19:14:19Z

Just an update, that I'm working on the minor changes requested above, but I'm getting golang crashes when building the docker image. make build works just fine, but make docker-build is not. I'm investigating what has recently changed, or if it's something on my end. EDIT: fixed. erase all / fresh install (turn it off then back on)

Signed-off-by: utdrmac <matthew.boehm@percona.com>

jdheyburn · 2026-04-14T19:42:48Z

I've deployed this locally, and works perfectly! @utdrmac there are some merge conflicts from the latest PR I merged - which was @hieu2102's PR adding operator and exporter user. If you clear these up then I think we can get this approved and merged.

bjosv

LGTM, just added a nit for the record.
Just a rebase needed, then open comments can be addressed in separate PR.

bjosv · 2026-04-17T13:52:50Z

 // buildValkeyNodeConfigMap builds a ConfigMap containing the embedded liveness
 // and readiness probe scripts, plus an empty valkey.conf.
-// The ConfigMap is named after valkeyNodeResourceName(node).
+// The ConfigMap is named via config.go:getConfigMapName(node).


nit: remove this line or change to GetServerConfigMapName, getConfigMapName was maybe used in a previous iteration.

Signed-off-by: utdrmac <matthew.boehm@percona.com>

utdrmac · 2026-04-20T23:56:39Z

@sandeepkunusoth Please check my merge with your latest TLS, as the config is moved into a separate file.

Signed-off-by: utdrmac <matthew.boehm@percona.com>

…eyConfig Signed-off-by: Joseph Heyburn <jdheyburn@gmail.com>

jdheyburn · 2026-04-22T11:17:48Z

@utdrmac I rebased this branch from main, looks like its causing some e2e tests to fail - would you mind taking a look?

utdrmac · 2026-04-24T14:42:57Z

🎉 Thanks everyone for helping, commenting, and testing!

utdrmac added 3 commits January 9, 2026 15:44

(feat): Extra Valkey Config

a05d5d6

This PR implements the ability for users to supply additional configuration for Valkey. Any provided parameters are appended to default configuration required by the operator Signed-off-by: utdrmac <matthew.boehm@percona.com>

corrected map name

7241a35

Signed-off-by: utdrmac <matthew.boehm@percona.com>

update sample cr

05af544

Signed-off-by: utdrmac <matthew.boehm@percona.com>

utdrmac mentioned this pull request Jan 9, 2026

(feat): Add support for auto-mounting extra configs #42

Closed

sandeepkunusoth reviewed Jan 9, 2026

View reviewed changes

Comment thread config/samples/v1alpha1_valkeycluster.yaml Outdated

Comment thread internal/controller/valkeycluster_controller.go

Comment thread internal/controller/valkeycluster_controller.go Outdated

utdrmac added 2 commits January 9, 2026 18:57

e2e fix; revert func name

8919932

Signed-off-by: utdrmac <matthew.boehm@percona.com>

config as map

f066a91

Signed-off-by: utdrmac <matthew.boehm@percona.com>

sandeepkunusoth reviewed Jan 10, 2026

View reviewed changes

Comment thread api/v1alpha1/valkeycluster_types.go

utdrmac added 2 commits January 9, 2026 20:34

add TODO comment, and e2e test

cc614b7

Signed-off-by: utdrmac <matthew.boehm@percona.com>

add TODO comment, and e2e test

88cf17a

Signed-off-by: utdrmac <matthew.boehm@percona.com>

sandeepkunusoth approved these changes Jan 10, 2026

View reviewed changes

sandeepkunusoth previously requested changes Jan 10, 2026

View reviewed changes

fix lint catch

02808a3

Signed-off-by: utdrmac <matthew.boehm@percona.com>

utdrmac requested a review from sandeepkunusoth January 10, 2026 03:04

e2e lint

59bcd2e

Signed-off-by: utdrmac <matthew.boehm@percona.com>

jdheyburn reviewed Jan 10, 2026

View reviewed changes

utdrmac added 2 commits January 12, 2026 15:36

merge main

3a1709a

Signed-off-by: utdrmac <matthew.boehm@percona.com>

merge upstream; expand e2e client commands

8dce649

Signed-off-by: utdrmac <matthew.boehm@percona.com>

sandeepkunusoth reviewed Jan 12, 2026

View reviewed changes

Comment thread api/v1alpha1/valkeycluster_types.go Outdated

bjosv previously approved these changes Jan 14, 2026

View reviewed changes

Merge branch 'main' into valkeyConfig

26e809e

Signed-off-by: utdrmac <matthew.boehm@percona.com>

sandeepkunusoth reviewed Jan 15, 2026

View reviewed changes

Comment thread internal/controller/valkeycluster_controller.go

utdrmac mentioned this pull request Jan 17, 2026

(feat): Support user creation on cluster init #56

Closed

refactor server config

f487a3c

Signed-off-by: utdrmac <matthew.boehm@percona.com>

merge upstream:main

91070a6

Signed-off-by: utdrmac <matthew.boehm@percona.com>

jdheyburn reviewed Apr 2, 2026

View reviewed changes

utdrmac added 4 commits April 6, 2026 20:05

Merge branch 'ghmain' into ghvalkeyconfig

9dccbf6

refactor building config

4af5fc0

Signed-off-by: utdrmac <matthew.boehm@percona.com>

update test

810b6bf

Signed-off-by: utdrmac <matthew.boehm@percona.com>

update tests; revert configmap name

519985d

Signed-off-by: utdrmac <matthew.boehm@percona.com>

utdrmac added 3 commits April 9, 2026 19:17

externalize config name helper

8c19a1a

Signed-off-by: utdrmac <matthew.boehm@percona.com>

Merge branch 'ghmain' into ghvalkeyconfig

14b4c8d

Merge branch 'ghmain' into ghvalkeyconfig

daaffd0

bjosv reviewed Apr 17, 2026

View reviewed changes

utdrmac mentioned this pull request Apr 17, 2026

[Design] Handling live configuration changes #141

Open

utdrmac added 3 commits April 20, 2026 18:49

merge main

b1c243a

Signed-off-by: utdrmac <matthew.boehm@percona.com>

gofmt

3c07b8b

Signed-off-by: utdrmac <matthew.boehm@percona.com>

merge main 2b3718a

d6587f7

Signed-off-by: utdrmac <matthew.boehm@percona.com>

sandeepkunusoth reviewed Apr 21, 2026

View reviewed changes

Comment thread internal/controller/config.go Outdated

move getTLSConfig back to util

60e9428

Signed-off-by: utdrmac <matthew.boehm@percona.com>

sandeepkunusoth reviewed Apr 21, 2026

View reviewed changes

Comment thread internal/controller/config.go Outdated

sandeepkunusoth reviewed Apr 21, 2026

View reviewed changes

Comment thread internal/controller/config.go

sandeepkunusoth reviewed Apr 21, 2026

View reviewed changes

Comment thread Makefile Outdated

sort base config keys

4a3409c

Signed-off-by: utdrmac <matthew.boehm@percona.com>

sandeepkunusoth approved these changes Apr 21, 2026

View reviewed changes

Merge branch 'main' of github.com:valkey-io/valkey-operator into valk…

e2dc8e6

…eyConfig Signed-off-by: Joseph Heyburn <jdheyburn@gmail.com>

jdheyburn force-pushed the valkeyConfig branch from d09b178 to e2dc8e6 Compare April 22, 2026 10:41

Merge branch 'valkey-io:main' into valkeyConfig

20fee93

sandeepkunusoth merged commit 4f3e468 into valkey-io:main Apr 23, 2026
8 checks passed

Conversation

utdrmac commented Jan 9, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

sandeepkunusoth left a comment

Choose a reason for hiding this comment

Uh oh!

sandeepkunusoth left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

utdrmac commented Jan 10, 2026

Uh oh!

jdheyburn left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

utdrmac commented Jan 12, 2026

Uh oh!

Uh oh!

sandeepkunusoth commented Jan 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

utdrmac commented Jan 12, 2026

Uh oh!

sandeepkunusoth commented Jan 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

bjosv left a comment

Choose a reason for hiding this comment

Uh oh!

bjosv commented Jan 15, 2026

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

utdrmac commented Apr 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jdheyburn commented Apr 14, 2026

Uh oh!

sandeepkunusoth left a comment •

edited

Loading

sandeepkunusoth commented Jan 12, 2026 •

edited

Loading

sandeepkunusoth commented Jan 12, 2026 •

edited

Loading

utdrmac commented Apr 9, 2026 •

edited

Loading