refactor: Extract apply and update status steps - #811
Open
adwk67 wants to merge 6 commits into
Open
Conversation
12 tasks
adwk67
commented
Jul 31, 2026
| // Discovery CM will fail to build until the rest of the cluster has been | ||
| // deployed, so do it last so that failure won't inhibit the rest of the | ||
| // cluster from booting up. | ||
| let discovery_cm = build_discovery_config_map( |
Member
Author
There was a problem hiding this comment.
This - build_discovery_config_map - is a build step, but it cannot live in the "actual" step (due to dependencies) but neither is it really part of apply, so I've left it here.
Member
Author
There was a problem hiding this comment.
This will be covered by our TODO item:
Address the race condition for discovery config maps: see HDFS, Hive, Kafka
Member
Author
There was a problem hiding this comment.
See: 24908e6
The listeners are not created by the operator but are needed to create the discovery config map. The listeners are watched so that complete listener data triggers a reconcile.
Member
Author
There was a problem hiding this comment.
Re-tested:
--- PASS: kuttl (1052.01s)
--- PASS: kuttl/harness (0.00s)
--- PASS: kuttl/harness/smoke_hadoop-3.5.0_zookeeper-3.9.4_zookeeper-latest-3.9.5_number-of-datanodes-2_datanode-pvcs-2hdd-1ssd_listener-class-cluster-internal_openshift-false (150.36s)
--- PASS: kuttl/harness/smoke_hadoop-3.5.0_zookeeper-3.9.5_zookeeper-latest-3.9.5_number-of-datanodes-2_datanode-pvcs-2hdd-1ssd_listener-class-cluster-internal_openshift-false (616.86s)
--- PASS: kuttl/harness/smoke_hadoop-3.5.0_zookeeper-3.9.5_zookeeper-latest-3.9.5_number-of-datanodes-2_datanode-pvcs-2hdd-1ssd_listener-class-external-unstable_openshift-false (140.38s)
--- PASS: kuttl/harness/smoke_hadoop-3.5.0_zookeeper-3.9.4_zookeeper-latest-3.9.5_number-of-datanodes-2_datanode-pvcs-2hdd-1ssd_listener-class-external-unstable_openshift-false (144.40s)
PASS
adwk67
commented
Jul 31, 2026
adwk67
marked this pull request as ready for review
July 31, 2026 11:08
Member
Author
|
siegfriedweber
self-requested a review
August 4, 2026 11:55
Member
Author
|
Openshift tests to verify clusterrole: |
siegfriedweber
requested changes
Aug 10, 2026
Comment on lines
+212
to
+233
| /// Applies the discovery `ConfigMap` directly, outside the [`ClusterResources`] tracking. | ||
| /// | ||
| /// The CM must stay untracked because it is not emitted on every reconcile run: the namenode | ||
| /// `Listener`s are per-pod, so whenever a namenode is scaled up (or a `Listener` briefly has | ||
| /// no ingress address), [`crate::crd::namenode_listener_refs`] returns `None` and the CM is | ||
| /// skipped for that run. A *tracked* CM would be deleted as an orphan by | ||
| /// [`ClusterResources::delete_orphaned_resources`] in every such window and re-created once | ||
| /// the addresses are back -- churn for every client watching the discovery CM. Untracked, the | ||
| /// existing CM simply stays in place until it can be rebuilt. | ||
| /// | ||
| /// This deliberately differs from Hive and Druid, which do track their discovery CMs: their | ||
| /// skip window only opens when a role `Listener` is deleted and re-created, whereas HDFS | ||
| /// would hit it on every namenode scale-up. | ||
| pub async fn apply_discovery_config_map(client: &Client, discovery_cm: &ConfigMap) -> Result<()> { | ||
| client | ||
| .apply_patch(FIELD_MANAGER_SCOPE, discovery_cm, discovery_cm) | ||
| .await | ||
| .with_context(|_| ApplyDiscoveryConfigMapSnafu { | ||
| name: discovery_cm.metadata.name.clone().unwrap_or_default(), | ||
| })?; | ||
| Ok(()) | ||
| } |
Member
There was a problem hiding this comment.
Should we re-emit the ConfigMap like the Druid operator does?
Comment on lines
+555
to
+559
| listener_name.starts_with(&format!( | ||
| "{listener_volume}-{cluster_name}-{role}-", | ||
| listener_volume = *LISTENER_VOLUME_NAME, | ||
| role = HdfsNodeRole::Name, | ||
| )) |
Member
There was a problem hiding this comment.
This implementation is good enough for now, but I want to note a possible name collision. Since a cluster name may itself contain -, that character doesn't reliably separate the parts of the listener name. For example, is_namenode_listener matches the namenode listeners of both the hdfs cluster and the hdfs-namenode cluster: the latter's namenode listeners are named listener-hdfs-namenode-namenode-<rolegroup>-<n>, which also starts with the listener-hdfs-namenode- prefix built for cluster hdfs.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
This PR covers the following:
Definition of Done Checklist
Author
Reviewer
Acceptance
type/deprecationlabel & add to the deprecation scheduletype/experimentallabel & add to the experimental features tracker