Add Stretch Cluster diagrams, 2/2/2 layout, and failure guidance#1820
Open
JakeSCahill wants to merge 9 commits into
Open
Add Stretch Cluster diagrams, 2/2/2 layout, and failure guidance#1820JakeSCahill wants to merge 9 commits into
JakeSCahill wants to merge 9 commits into
Conversation
Chooser page: - Add a topology comparison diagram (multi-AZ, Stretch Cluster, Shadowing) above the comparison table - Add a Stretch Cluster vs Shadowing comparison table (replication, RPO, RTO, client failover, topology, produce latency, best for) - Add Tiered Storage and relative cost rows to the topology table, concrete produce-latency figures, and a Whole Cluster Restore pointer Stretch page: - Add an architecture diagram separating the operator control plane from the broker data plane, with split-brain protection and two-level leader election explained - Recommend a 2/2/2 six-broker layout (survives location loss and restores full RF=5 replication after a single broker failure); validated end-to-end on kind - Add a Limitations section: no Tiered Storage (and no Iceberg topics or Cloud Topics), same-continent placement with example region trios - Add an Understand failure scenarios section (one location down, two or more down, beyond quorum with Shadowing as defense in depth) - Document spec-drift behavior (SpecSynced condition, DriftDetected) with a troubleshooting entry, the fixed peer list and how to change it, managed EKS/GKE/AKS support, plugin auto-download, kubeconfig caching, ordered_racks leader preference with follower fetching, and the operator's automatic stale-disk wipe recovery
✅ Deploy Preview for redpanda-docs-preview ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
Contributor
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Per engineering (K8S-907 discussion): in this release the RedpandaBrokerPool CRD is installed only in multicluster mode and its controller reconciles pools only for StretchCluster resources; NodePool (beta) remains the pool mechanism for standalone clusters. - Replace the release-notes section that claimed NodePool documentation was removed (the NodePool docs are still published) with an accurate 'Broker pool resources by deployment type' section - Add a Limitations bullet on the stretch page pointing standalone deployments at NodePool Verified against operator source at operator/v26.2.1-beta.3: multiclusterCRDs gating in cmd/crd/crd.go and the StretchCluster-only clusterRef resolution in brokerpool_controller.go
- Fix the Cloud Topics xref, which moved under cloud-topics/ on beta - Spell out RPO on first use on the stretch page - Rework the Limitations bullets to the page's bold-term-colon pattern - Tie the 50 ms recommendation to the 100 ms liveness ceiling in one sentence and drop a doubled example marker - Replace 'as well as' with 'and'; use active voice for the peer-list sentence - Chooser: parallel phrasing in the Relative cost row, clearer Whole Cluster Restore trade-off sentence, table spacing consistency
Set flowchart.subGraphTitleMargin (bottom: 16) in the Mermaid initialize options so subgraph labels no longer sit directly on the nested boxes below them. The top margin must stay 0 because Mermaid 10 shifts the label up by the top margin without growing the subgraph box, which clips the title.
Mermaid 10 and 11 never reserve vertical space for a cluster label above nested subgraphs: the label overlaps the inner box borders, and neither flowchart.subGraphTitleMargin nor flowchart.padding changes the label-to-content distance (measured at -3px for any value). The only working lever is a trailing blank line in the title itself.
…y diagram" This reverts commit d3599d8.
This reverts commit ccb582e.
Contributor
|
We should be suggestion 2/2/1 not 2/2/2 so that if one region goes down we always have quorum due to an odd number of replicas also we need to recommend RF=5 so all replicas have copies of data in case customers want to preserve data even if the quorum is lost. Also the digram looks a bit hard to read is possible to arrange the clusters in a triangle pattern to show cross cluster communication instead of having them stacked on each other? |
…side Per review: 2/2/1 with default_topic_replications: 5 keeps an odd broker count, so the cluster always holds a clean three-of-five majority through the loss of any one location, and with five replicas on five brokers every broker holds a copy of every partition, so any surviving location retains a complete copy of the data even if quorum is lost. Also lay the architecture diagram out left-to-right so the three Kubernetes clusters appear side by side with the full cross-cluster mesh visible, instead of stacked vertically.
Contributor
Author
Preview rebuilds from the latest push. |
Contributor
|
Looks good but I'm not sure if its just me there are extraneous plus signs in the doc.
|
david-yu
approved these changes
Jul 24, 2026
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.
Follow-up to #1681, which merged while this work was in flight.
Chooser page (
k-choose-deployment.adoc)Stretch page (
k-stretch-clusters.adoc)rpk cluster info, RF=5 topic)SpecSyncedcondition, reasonDriftDetected, reconciliation blocks but pod remediation continues) with a troubleshooting entryrpk k8suse; kubeconfig caching for instant leader takeover;ordered_racksleader preference paired with follower fetching; automatic stale-disk wipe recovery (--wipe-stale-disk-after)Notes for reviewers
k8s-api-...-stretchclusterand...-redpandabrokerpoolink-crd.adoc): they resolve once docs: ShadowLink schema replication and role sync on Kubernetes (DOC-2333) #1817 merges, which regenerates the CRD reference with the stretch types. docs: ShadowLink schema replication and role sync on Kubernetes (DOC-2333) #1817 should merge before or with this PR.operator/v26.2.1-beta.3(spec-drift condition names, stale-disk wipe flag semantics, local Lease election flags) and against eng guidance in Slack (Tiered Storage, same-continent, EKS/GKE/AKS, peer-list changes).🤖 Generated with Claude Code