feat: add disaster recovery solution documentation for GitLab #80
Conversation
Signed-off-by: mingfu <[email protected]>
WalkthroughAdds six new disaster-recovery solution documents (English and Chinese) for GitLab, SonarQube, and Nexus describing hot-data/cold-compute architectures, data replication mechanisms (Ceph RBD, PostgreSQL streaming, object replication), failover/runbook procedures, setup steps, backup guidance, and RPO/RTO targets. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
participant User
participant Primary_App as Primary App (GitLab/SonarQube/Nexus)
participant Primary_DB as Primary DB (Postgres)
participant Object_Storage as Object Storage (S3/Ceph)
participant Block_Storage as Block Storage (Ceph RBD)
participant Replica_DB as DR DB (standby)
participant Secondary_App as Secondary (cold compute)
rect rgb(230,248,230)
Note over Primary_DB,Replica_DB: Continuous DB replication / WAL streaming
end
rect rgb(230,248,230)
Note over Block_Storage: Block-level replication (RBD Mirror) / PV snapshots
end
User->>Primary_App: Normal read/write traffic
Primary_App->>Primary_DB: write transactions
Primary_App->>Object_Storage: upload objects/attachments
Primary_App->>Block_Storage: PV writes (Gitaly / repo data)
Primary_DB-->>Replica_DB: WAL stream / async replication
Object_Storage-->>Object_Storage: bucket/object replication
Block_Storage-->>Block_Storage: RBD mirror replication
Primary_App->>Secondary_App: periodic backups / PV snapshots (cold)
sequenceDiagram
autonumber
participant Operator
participant Replica_DB as Promote DB
participant Secondary_App as Activate Secondary
participant DNS
participant User
rect rgb(255,244,230)
Note over Operator,Replica_DB: Failover runbook (promote DB, restore PVs)
end
Operator->>Replica_DB: promote standby to primary
Operator->>Secondary_App: restore/attach PVs, apply secrets, start services
Operator->>DNS: update external routing / load balancer
DNS-->>User: traffic routed to Secondary_App
Secondary_App->>Replica_DB: accept writes (system active)
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes
Suggested reviewers
Poem
Pre-merge checks and finishing touches✅ Passed checks (3 passed)
✨ Finishing touches🧪 Generate unit tests (beta)
Tip 📝 Customizable high-level summaries are now available in beta!You can now customize how CodeRabbit generates the high-level summary in your pull requests — including its content, structure, tone, and formatting.
Example instruction:
Note: This feature is currently in beta for Pro-tier users, and pricing will be announced later. 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 |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (2)
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md (2)
344-344: Use hyphenated compound adjective form for "backed-up" before nouns.When "backed up" modifies a noun that follows, it should be hyphenated as "backed-up" per standard English grammar. This affects multiple locations: lines 344, 350, 395, 397, 462, and 477.
Apply these diffs to correct the compound adjectives:
- pg-secret.yaml: Change the `host` and `password` fields to the PostgreSQL connection address and password of the secondary cluster + pg-secret.yaml: Change the `host` and `password` fields to the secondary cluster's PostgreSQL connection address and passwordAlternatively, apply the hyphenation directly at each occurrence:
- Modify the backed up files: + Modify the backed-up files:- Create the backed up YAML files in the disaster recovery environment + Create the backed-up YAML files in the disaster recovery environment- Modify the three backed up PV files + Modify the three backed-up PV files- Restore the backed up PVC and PV resources + Restore the backed-up PVC and PV resources- Restore the backed up `gitlabofficial.yaml` + Restore the backed-up `gitlabofficial.yaml`Also applies to: 350-350, 395-395, 397-397, 462-462, 477-477
115-117: Reduce repetition in prerequisite sentences.Lines 115–117 begin with identical phrasing ("Complete the deployment of..."). Rewording for variety improves readability.
- 2. Complete the deployment of `Alauda support for PostgreSQL` disaster recovery configuration. - 3. Complete the deployment of `Alauda Build of Rook-Ceph` object storage disaster recovery configuration ([optional if conditions are met](#data-synchronization-strategy)). - 4. Complete the deployment of `Alauda Build of Rook-Ceph` block storage disaster recovery configuration. + 2. Deploy `Alauda support for PostgreSQL` disaster recovery configuration. + 3. Set up `Alauda Build of Rook-Ceph` object storage disaster recovery configuration ([optional if conditions are met](#data-synchronization-strategy)). + 4. Configure `Alauda Build of Rook-Ceph` block storage disaster recovery.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
⛔ Files ignored due to path filters (1)
docs/public/gitlab-disaster-recovery.drawio.svgis excluded by!**/*.svg
📒 Files selected for processing (2)
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md(1 hunks)docs/zh/solutions/How_to_perform_disaster_recovery_for_gitlab.md(1 hunks)
🧰 Additional context used
🪛 LanguageTool
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md
[style] ~65-~65: Consider an alternative for the often overused word ‘important’.
Context: ...ge. If you assess that this data is not important, you can choose not to perform disaster...
(NOT_IMPORTANT)
[style] ~117-~117: Three successive sentences begin with the same word. Consider rewording the sentence or use a thesaurus to find a synonym.
Context: ...t](#data-synchronization-strategy)). 4. Complete the deployment of `Alauda Build of Rook...
(ENGLISH_WORD_REPEAT_BEGINNING_RULE)
[grammar] ~344-~344: Use a hyphen to join words.
Context: ...he following modifications to the backed up files: - pg-secret.yaml: Change the ...
(QB_NEW_EN_HYPHEN)
[grammar] ~350-~350: Use a hyphen to join words.
Context: ...the secondary cluster Create the backed up YAML files in the disaster recovery e...
(QB_NEW_EN_HYPHEN)
[grammar] ~395-~395: Use a hyphen to join words.
Context: ...cho "" done ``` Modify the three backed up PV files and delete all `spec.claimRe...
(QB_NEW_EN_HYPHEN)
[grammar] ~397-~397: Use a hyphen to join words.
Context: ...` fields in the yaml. Create the backed up PVC and PV YAML files directly in the...
(QB_NEW_EN_HYPHEN)
[style] ~412-~412: You have already used this phrasing in nearby sentences. Consider replacing it to add variety to your writing.
Context: ...rills. After the drill is complete, you need to perform the following cleanup operation...
(REP_NEED_TO_VB)
[grammar] ~462-~462: Use a hyphen to join words.
Context: ...C and PV Resources**: Restore the backed up PVC and PV resources to the disaster ...
(QB_NEW_EN_HYPHEN)
[grammar] ~477-~477: Use a hyphen to join words.
Context: ...y Secondary GitLab**: Restore the backed up gitlabofficial.yaml to the disaster...
(QB_NEW_EN_HYPHEN)
docs/zh/solutions/How_to_perform_disaster_recovery_for_gitlab.md
[uncategorized] ~108-~108: 动词的修饰一般为‘形容词（副词）+地+动词’。您的意思是否是：一致"地"命名
Context: ... Ceph 存储池名称和存储类名称 - 一致的 GitLab 实例名称 - 一致的命名空间名称 ::: ### 前置条件 1. 提前准备一个主集群和一个灾难...
(wb4)
🪛 markdownlint-cli2 (0.18.1)
docs/zh/solutions/How_to_perform_disaster_recovery_for_gitlab.md
412-412: Link fragments should be valid
(MD051, link-fragments)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: build
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md
Outdated
Show resolved
Hide resolved
docs/zh/solutions/How_to_perform_disaster_recovery_for_gitlab.md
Outdated
Show resolved
Hide resolved
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md (1)
344-344: Use hyphenated compound adjectives for "backed-up".When "backed up" functions as a compound adjective modifying a noun, it should be hyphenated as "backed-up" for proper English grammar. Review and update the following phrases:
- Line 344: "backed up files" → "backed-up files"
- Line 350: "backed up YAML files" → "backed-up YAML files"
- Line 395: "backed up PV files" → "backed-up PV files"
- Line 397: "backed up PVC and PV YAML files" → "backed-up PVC and PV YAML files"
- Line 462: "backed up PVC and PV resources" → "backed-up PVC and PV resources"
- Line 477: "backed up
gitlabofficial.yaml" → "backed-upgitlabofficial.yaml"Also applies to: 350-350, 395-395, 397-397, 462-462, 477-477
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (4)
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md(1 hunks)docs/en/solutions/How_to_perform_disaster_recovery_for_sonarqube.md(1 hunks)docs/zh/solutions/How_to_perform_disaster_recovery_for_gitlab.md(1 hunks)docs/zh/solutions/How_to_perform_disaster_recovery_for_sonarqube.md(1 hunks)
🧰 Additional context used
🪛 LanguageTool
docs/zh/solutions/How_to_perform_disaster_recovery_for_sonarqube.md
[uncategorized] ~72-~72: 动词的修饰一般为‘形容词（副词）+地+动词’。您的意思是否是：一致"地"命名
Context: ... 一致的数据库实例名称和密码 - 一致的 SonarQube 实例名称 - 一致的命名空间名称 ::: ### 前置条件 1. 提前准备一个主集群和一个灾难...
(wb4)
docs/zh/solutions/How_to_perform_disaster_recovery_for_gitlab.md
[uncategorized] ~108-~108: 动词的修饰一般为‘形容词（副词）+地+动词’。您的意思是否是：一致"地"命名
Context: ... Ceph 存储池名称和存储类名称 - 一致的 GitLab 实例名称 - 一致的命名空间名称 ::: ### 前置条件 1. 提前准备一个主集群和一个灾难...
(wb4)
docs/en/solutions/How_to_perform_disaster_recovery_for_sonarqube.md
[style] ~175-~175: You have already used this phrasing in nearby sentences. Consider replacing it to add variety to your writing.
Context: ...rills. After the drill is complete, you need to perform the following cleanup operation...
(REP_NEED_TO_VB)
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md
[style] ~65-~65: Consider an alternative for the often overused word ‘important’.
Context: ...ge. If you assess that this data is not important, you can choose not to perform disaster...
(NOT_IMPORTANT)
[style] ~117-~117: Three successive sentences begin with the same word. Consider rewording the sentence or use a thesaurus to find a synonym.
Context: ...t](#data-synchronization-strategy)). 4. Complete the deployment of `Alauda Build of Rook...
(ENGLISH_WORD_REPEAT_BEGINNING_RULE)
[grammar] ~344-~344: Use a hyphen to join words.
Context: ...he following modifications to the backed up files: - pg-secret.yaml: Change the ...
(QB_NEW_EN_HYPHEN)
[grammar] ~350-~350: Use a hyphen to join words.
Context: ...the secondary cluster Create the backed up YAML files in the disaster recovery e...
(QB_NEW_EN_HYPHEN)
[grammar] ~395-~395: Use a hyphen to join words.
Context: ...cho "" done ``` Modify the three backed up PV files and delete all `spec.claimRe...
(QB_NEW_EN_HYPHEN)
[grammar] ~397-~397: Use a hyphen to join words.
Context: ...` fields in the yaml. Create the backed up PVC and PV YAML files directly in the...
(QB_NEW_EN_HYPHEN)
[style] ~412-~412: You have already used this phrasing in nearby sentences. Consider replacing it to add variety to your writing.
Context: ...rills. After the drill is complete, you need to perform the following cleanup operation...
(REP_NEED_TO_VB)
[grammar] ~462-~462: Use a hyphen to join words.
Context: ...C and PV Resources**: Restore the backed up PVC and PV resources to the disaster ...
(QB_NEW_EN_HYPHEN)
[grammar] ~477-~477: Use a hyphen to join words.
Context: ...y Secondary GitLab**: Restore the backed up gitlabofficial.yaml to the disaster...
(QB_NEW_EN_HYPHEN)
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (2)
docs/en/solutions/How_to_perform_disaster_recovery_for_nexus.md (1)
236-236: Apply consistent hyphenation for compound adjectives with "backed up".When "backed up" modifies a noun (compound adjective), it should be hyphenated: "backed-up PV file", "backed-up PVC and PV YAML files", etc. This improves grammatical clarity.
Apply this diff to fix the hyphenation:
- Modify the backed up PV file and delete all `spec.claimRef` fields in the yaml. + Modify the backed-up PV file and delete all `spec.claimRef` fields in the yaml.- Create the backed up PVC and PV YAML files directly in the disaster recovery environment with the same namespace name. + Create the backed-up PVC and PV YAML files directly in the disaster recovery environment with the same namespace name.- Restore the backed up PVC and PV resources to the disaster recovery environment with the same namespace name, and check that the PVC status in the secondary cluster is `Bound`: + Restore the backed-up PVC and PV resources to the disaster recovery environment with the same namespace name, and check that the PVC status in the secondary cluster is `Bound`:- 4. **Deploy Secondary Nexus**: Restore the backed up `nexus.yaml` to the disaster recovery environment with the same namespace name. Nexus will automatically start using the disaster recovery data. + 4. **Deploy Secondary Nexus**: Restore the backed-up `nexus.yaml` to the disaster recovery environment with the same namespace name. Nexus will automatically start using the disaster recovery data.Also applies to: 238-238, 289-289, 300-300
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md (1)
395-395: Apply consistent hyphenation for compound adjectives with "backed up".Similar to the Nexus documentation, when "backed up" modifies a noun, it should be hyphenated for grammatical consistency. Line 395 should use "backed-up" in compound adjective form.
Apply this diff:
- Modify the three backed up PV files and delete all `spec.claimRef` fields in the yaml. + Modify the three backed-up PV files and delete all `spec.claimRef` fields in the yaml.Also applies to: 397-397
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (4)
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md(1 hunks)docs/en/solutions/How_to_perform_disaster_recovery_for_nexus.md(1 hunks)docs/zh/solutions/How_to_perform_disaster_recovery_for_gitlab.md(1 hunks)docs/zh/solutions/How_to_perform_disaster_recovery_for_nexus.md(1 hunks)
🧰 Additional context used
🪛 LanguageTool
docs/zh/solutions/How_to_perform_disaster_recovery_for_nexus.md
[uncategorized] ~73-~73: 动词的修饰一般为‘形容词（副词）+地+动词’。您的意思是否是：一致"地"命名
Context: ...的 Ceph 存储池名称和存储类名称 - 一致的 Nexus 实例名称 - 一致的命名空间名称 ::: ### 前置条件 1. 提前准备一个主集群和一个灾难...
(wb4)
docs/zh/solutions/How_to_perform_disaster_recovery_for_gitlab.md
[uncategorized] ~108-~108: 动词的修饰一般为‘形容词（副词）+地+动词’。您的意思是否是：一致"地"命名
Context: ... Ceph 存储池名称和存储类名称 - 一致的 GitLab 实例名称 - 一致的命名空间名称 ::: ### 前置条件 1. 提前准备一个主集群和一个灾难...
(wb4)
docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md
[style] ~65-~65: Consider an alternative for the often overused word ‘important’.
Context: ...ge. If you assess that this data is not important, you can choose not to perform disaster...
(NOT_IMPORTANT)
[style] ~117-~117: Three successive sentences begin with the same word. Consider rewording the sentence or use a thesaurus to find a synonym.
Context: ...t](#data-synchronization-strategy)). 4. Complete the deployment of `Alauda Build of Rook...
(ENGLISH_WORD_REPEAT_BEGINNING_RULE)
[style] ~412-~412: You have already used this phrasing in nearby sentences. Consider replacing it to add variety to your writing.
Context: ...rills. After the drill is complete, you need to perform the following cleanup operation...
(REP_NEED_TO_VB)
docs/en/solutions/How_to_perform_disaster_recovery_for_nexus.md
[grammar] ~236-~236: Use a hyphen to join words.
Context: ...pv-${PV}.yaml" fi ``` Modify the backed up PV file and delete all `spec.claimRef...
(QB_NEW_EN_HYPHEN)
[grammar] ~238-~238: Use a hyphen to join words.
Context: ...` fields in the yaml. Create the backed up PVC and PV YAML files directly in the...
(QB_NEW_EN_HYPHEN)
[style] ~253-~253: You have already used this phrasing in nearby sentences. Consider replacing it to add variety to your writing.
Context: ...rills. After the drill is complete, you need to perform the following cleanup operation...
(REP_NEED_TO_VB)
[grammar] ~289-~289: Use a hyphen to join words.
Context: ...C and PV Resources**: Restore the backed up PVC and PV resources to the disaster ...
(QB_NEW_EN_HYPHEN)
[grammar] ~300-~300: Use a hyphen to join words.
Context: ...oy Secondary Nexus**: Restore the backed up nexus.yaml to the disaster recovery...
(QB_NEW_EN_HYPHEN)
🔇 Additional comments (18)
docs/en/solutions/How_to_perform_disaster_recovery_for_nexus.md (1)
115-135: Ensure consistency in storage class naming conventions.Line 110 references
ceph-rdbas the storage class name with a comment suggesting it should be "the configured storage class name." This is appropriately flexible, and the documentation later provides commands to verify the setup. However, ensure this naming convention is consistently documented across the Nexus, GitLab, and other DR solutions. The naming appears consistent with the GitLab DR documentation, so this is acceptable.docs/zh/solutions/How_to_perform_disaster_recovery_for_gitlab.md (2)
412-412: Verify markdown link fragment matches target heading.Line 412 references the internal link
[灾难切换](#灾难切换)which correctly targets the section heading## 灾难切换at line 452. This appears to have been properly fixed from the previous review. The link should work correctly in rendered documentation.Also applies to: 452-452
41-41: Diagram file reference is valid.The file
public/gitlab-disaster-recovery.drawio.svgexists in the repository. The relative path../../public/gitlab-disaster-recovery.drawio.svgfrom the Chinese documentation location correctly resolves to the diagram file.docs/en/solutions/How_to_perform_disaster_recovery_for_gitlab.md (2)
412-412: Verify markdown link fragment matches target heading.Line 412 references the internal link
[Primary-Secondary Switchover Procedure in Disaster Scenarios](#primary-secondary-switchover-procedure-in-disaster-scenarios)which correctly targets the section heading## Primary-Secondary Switchover Procedure in Disaster Scenariosat line 452. The link fragment will correctly resolve when rendered.Also applies to: 452-452
41-41: Diagram file reference is valid and the file exists.The diagram file
gitlab-disaster-recovery.drawio.svgexists atdocs/public/gitlab-disaster-recovery.drawio.svg, which correctly resolves from the relative path../../public/gitlab-disaster-recovery.drawio.svgreferenced in line 41. No action required.docs/zh/solutions/How_to_perform_disaster_recovery_for_nexus.md (13)
1-16: Well-structured metadata and clear problem statement.The front matter is properly formatted, and the problem section clearly communicates the architecture approach and scope. The note that users must implement their own address-switching mechanism appropriately sets expectations.
22-33: Comprehensive and well-organized terminology section.All key concepts (RPO, RTO, failover, etc.) are pre-defined before use, making the document accessible to operators less familiar with DR terminology.
35-64: Clear architecture explanation with well-defined components.The hot-data/cold-compute pattern and five-step failover procedure are clearly explained. The approach of pre-deploying Nexus with zero replicas in the standby cluster is sound.
65-85: Sound prerequisites with practical consistency recommendations.The guidance to maintain consistent naming and pool names across environments significantly simplifies failover procedures. The prerequisite list is realistic and appropriately references external Ceph DR documentation.
261-282: Realistic recovery objectives with transparent component breakdown.The RPO correctly identifies dependency on Ceph RBD sync intervals and the RTO appropriately acknowledges manual procedures with realistic time estimates. The DNS propagation caveat for external routing is a thoughtful detail that operators should consider.
283-305: Clear and sequential disaster switchover procedure with verification steps.The six-step failover sequence is logical and includes appropriate status checks. The example output showing PVC in "Bound" state provides operators with a concrete success criterion. Deferring external address switch to the final step is the correct approach.
306-312: Appropriate scope disclaimer with production readiness caution.The warning to verify DR capability of alternative block storage and conduct testing before production use is sound operational guidance.
68-75: Grammar suggestion from static analysis appears to be a false positive.The static analysis tool suggested using "一致地命名" (naming consistently) instead of "一致的 ... 名称" (consistent ... names), but the current phrasing is grammatically correct. The context is a list of noun phrases describing attributes (consistent naming, not the act of naming), so using "一致的" as an adjective is appropriate.
96-111: YAML structure appears valid, but verify storage class naming.The Nexus YAML example is syntactically correct. However, line 110 shows
storageClass: name: ceph-rdb— note whether this should beceph-rbd(RBD = RADOS Block Device). This may be environment-specific, but the common Ceph RBD storage class naming convention uses "rbd" not "rdb". Please verify this matches your actual storage class names.
115-170: Command examples use standard kubectl and Ceph syntax with realistic output samples.The bash and kubectl commands follow standard conventions. The output examples (e.g., VolumeReplication status at lines 141-143 and rbd mirror image status at lines 157-170) are realistic and would help operators verify correct setup.
Confirm that the Ceph RBD Mirror command syntax (
rbd mirror snapshot schedule ls,rbd mirror image status) and the expected output formats match the actual tools available in the target environment (Rook-Ceph version 4.x).
1-312: Comprehensive and well-structured disaster recovery documentation with strong operational guidance.This documentation provides a complete runbook for Nexus disaster recovery using Ceph RBD Mirror replication. Strengths include:
- Clear architecture explanation with terminology glossary
- Defensive warnings preventing common mistakes (premature Nexus creation, correct PV backup procedures)
- Realistic RPO/RTO targets with transparent component breakdown
- Sequential, actionable failover procedures with concrete status checks
- Helpful distinction between manual procedures and their typical time costs
The document aligns well with the PR objective to provide disaster recovery guidance for multiple services (mentioned alongside GitLab and SonarQube).
Please verify the following before merging:
- External documentation links (docs.alauda.io/container_platform/4.1/...) are accessible and current
- Storage class naming in line 110 (
ceph-rdb) is correct for your environment (vs. typicalceph-rbd)- Ceph RBD Mirror command syntax and expected output formats match your actual Rook-Ceph/Ceph version
- The PV backup procedure (removing
spec.claimRefat line 236) aligns with Kubernetes PV lifecycle best practices
83-83: All external documentation links are currently accessible and functional.Verification testing confirms all four referenced links to
docs.alauda.ioreturn HTTP 200 status codes:
#create-volumereplicationclass- ✓ accessible- Base dr_block.html page - ✓ accessible
#enable-mirror-for-pvc- ✓ accessible#procedures-1- ✓ accessibleThe documentation URLs and anchor fragments are valid and pointing to existing content. No broken links or inaccessible endpoints were found.
86-171: Documentation commands and output formats are verified as accurate.Both Ceph RBD Mirror commands reference correct syntax and output formats:
rbd mirror snapshot schedule ls --pool $CEPH_BLOCK_POOL --recursive— Command syntax matches official Ceph documentation; output columns (POOL, NAMESPACE, IMAGE, SCHEDULE) and format examples ("every 1m") are accurate.
rbd mirror image status $CEPH_BLOCK_POOL/$NEXUS_BLOCK_IMAGE_NAME— Command syntax and output structure (global_id, state, description, service, last_update, peer_sites) align with official Ceph documentation. State values ("up+stopped", "up+replaying") are documented and correct. The nested JSON in the description field reflects real Ceph behavior.The practical setup instructions, YAML examples, and status-check outputs are technically sound and consistent with the Alauda Container Platform documentation pattern.
| ### 设置备用 Nexus | ||
|
|
||
| :::warning | ||
| 当 Ceph RBD 处于备用状态时，同步过来的存储块无法挂载，因此备集群的 Nexus 无法部署成功。 | ||
|
|
||
| 如需验证备集群 Nexus 是否可以部署成功，可以临时将备集群的 Ceph RBD 提升为主集群，测试完成后再设置回备用状态。同时需要将测试过程中创建的 Nexus、PV 和 PVC 资源都删除。 | ||
| ::: | ||
|
|
||
| 1. 备份主 Nexus 使用的 Secret | ||
| 2. 备份主集群 Nexus 组件的 PVC 和 PV 资源 YAML | ||
| 3. 备份主集群 Nexus 的 Nexus 资源 YAML | ||
|
|
||
| #### 备份主 Nexus 使用的 Secret | ||
|
|
||
| 获取主 Nexus 使用的 Password Secret YAML，并将 Secret 创建到备集群同名命名空间中。 | ||
|
|
||
| ```bash | ||
| apiVersion: v1 | ||
| data: | ||
| password: xxxxxx | ||
| kind: Secret | ||
| metadata: | ||
| name: nexus-root-password | ||
| namespace: nexus-dr | ||
| type: Opaque | ||
| ``` | ||
|
|
||
| #### 备份主 Nexus 组件的 PVC 和 PV 资源 | ||
|
|
||
| :::tip | ||
| PV 资源中保存了 volume 属性信息，这些信息是容灾恢复时的关键信息，需要备份好。 | ||
|
|
||
| ```bash | ||
| volumeAttributes: | ||
| clusterID: rook-ceph | ||
| imageFeatures: layering | ||
| imageFormat: "2" | ||
| imageName: csi-vol-459e6f28-a158-4ae9-b5da-163448c35119 | ||
| journalPool: myblock | ||
| pool: myblock | ||
| storage.kubernetes.io/csiProvisionerIdentity: 1763446982673-7963-rook-ceph.rbd.csi.ceph.com | ||
| ``` | ||
|
|
||
| ::: | ||
|
|
||
| 执行以下命令将主 Nexus 组件的 PVC 和 PV 资源备份到当前目录： | ||
|
|
||
| ```bash | ||
| export NEXUS_PVC_NAME=<PVC_NAME> | ||
|
|
||
| echo "=> Exporting PVC $NEXUS_PVC_NAME" | ||
|
|
||
| # 导出 PVC | ||
| kubectl -n "$NEXUS_NAMESPACE" get pvc "$NEXUS_PVC_NAME" -o yaml > "pvc-${NEXUS_PVC_NAME}.yaml" | ||
|
|
||
| # 获取 PV | ||
| PV=$(kubectl -n "$NEXUS_NAMESPACE" get pvc "$NEXUS_PVC_NAME" -o jsonpath='{.spec.volumeName}') | ||
|
|
||
| if [[ -n "$PV" ]]; then | ||
| echo " ↳ Exporting PV $PV" | ||
| kubectl get pv "$PV" -o yaml > "pv-${PV}.yaml" | ||
| fi | ||
| ``` | ||
|
|
||
| 修改备份出来的 PV 文件，将 yaml 中的 `spec.claimRef` 字段全部删除。 | ||
|
|
||
| 将备份出来的 PVC 和 PV YAML 文件直接创建到容灾环境同名命名空间中。 | ||
|
|
||
| #### 备份主 Nexus 实例 YAML | ||
|
|
||
| ```bash | ||
| kubectl -n "$NEXUS_NAMESPACE" get nexus "$NEXUS_NAME" -oyaml > nexus.yaml | ||
| ``` | ||
|
|
||
| 根据容灾环境实际情况修改 `nexus.yaml` 中的信息。 | ||
|
|
||
| :::warning | ||
| `Nexus` 资源**不需要**立即创建在容灾环境，只需要在灾难发生时，执行容灾切换时创建到备集群即可。 | ||
| ::: | ||
|
|
||
| :::warning | ||
| 如需进行容灾演练，可以按照 [灾难切换](#灾难切换) 中的步骤进行演练。演练完毕后需要在容灾环境完成以下清理操作： | ||
|
|
||
| - 将容灾环境中的 `Nexus` 实例删除 | ||
| - 将创建的 PVC 和 PV 删除 | ||
| - 将 Ceph RBD 切换为备用状态 | ||
|
|
||
| ::: | ||
|
|
There was a problem hiding this comment.
🧩 Analysis chain
Kubernetes PersistentVolume claimRef removal reuse across clusters procedure
<web_search_result>
Short procedure to remove a PVC claimRef and reuse the underlying volume (including across clusters)
Important notes up front
- Always snapshot/backup the underlying volume before you touch PV/PVC metadata. Cloud disks and CSI details vary and you can lose data. [1][2]
- Dynamic volumes with reclaimPolicy: Delete will remove the cloud disk when the PVC is deleted — change to Retain first. [1][4]
Steps (cluster A → cluster B)
- Snapshot/backup the cloud volume (provider console or CSI snapshot). [1]
- In cluster A: change the PV reclaimPolicy to Retain so Kubernetes will not delete the backing disk.
kubectl patch pv -p '{"spec":{"persistentVolumeReclaimPolicy":"Retain"}}' [3][4] - Ensure the pod using the PVC is stopped and the PVC is deleted (or will be deleted). Wait until the PV status becomes Released. [1][5]
- Remove the claimRef from the PV metadata (or at minimum remove the uid field). Two common approaches:
- Edit the PV and remove the entire spec.claimRef block: kubectl edit pv — remove claimRef section. [3][4]
- Safer: edit and only remove the uid/resourceVersion lines inside claimRef so a new PVC in the same namespace can bind predictably. Some docs recommend removing only uid. [3][4]
- At this point the PV should become Available (or you can recreate a PV manifest that points to the same underlying disk but without claimRef). If you plan to move across clusters, you usually must detach the cloud disk and create a new PV in cluster B that references the same disk (provider-specific: GCE PD, AWS EBS, Azure Disk, or CSI volumeHandle). Example: create a PV manifest in cluster B with the correct volume identifier or CSI volumeHandle and desired access mode/storageClass. [1][6]
- In cluster B: apply the PV (or let dynamic provisioning re-create if using a snapshot + new PVC). Then create a matching PVC in cluster B (same size/accessModes/storageClass) so it binds to the PV. Verify Bound. [1][5]
Additional cautions and tips
- Many providers require detaching the disk from cluster A nodes before attaching to cluster B; follow provider instructions (GCP/AWS/Azure). [1]
- If the PV was managed by a CSI driver, prefer using CSI snapshot/restore or provider tooling to move volumes rather than manual claimRef edits. CSI volumeHandle values are required when creating a PV that points to an existing volume. [1][6]
- If you find a finalizer blocking PVC deletion (pvc-protection), remove it only after you are certain the PV reclaim policy is Retain and you have a backup. [4]
References
[1] Kubernetes/GKE docs on PersistentVolumes and reclaiming volumes.
[2] Provider guidance: snapshot/backup before operations (GKE/AWS/Azure docs).
[3] StackOverflow / community steps to patch PV to Retain and remove claimRef.
[4] Kubernetes/OKD docs on manual reclaim and reclaimPolicy semantics.
[5] Alibaba Cloud ACK docs & migration notes (example of removing claimRef and setting Retain).
[6] CSI/provider migration recommendations for creating PVs that point to existing volumes.
If you want, tell me which cloud provider and I’ll give exact commands/manifests for that provider.
</web_search_result>
The documented PV backup and recovery procedure is incomplete; critical operational steps are missing that could result in data loss.
While removing spec.claimRef is the correct approach for reusing PVs across clusters, the documentation omits several essential steps:
- Missing reclaimPolicy change — Before removing the PVC, the PV's
reclaimPolicyshould be set toRetainto prevent the backing volume from being deleted - No status verification — The procedure should wait for the PV status to transition to
Releasedbefore reusing it - CSI volumeHandle considerations — For Ceph RBD CSI volumes, the
volumeHandlein the PV spec needs to be preserved or correctly referenced in the new cluster; simply removingclaimRefis insufficient without this context
The section should explicitly document:
- Patch the PV to set
persistentVolumeReclaimPolicy: Retainbefore PVC deletion - Verify the PV reaches
Releasedstatus after PVC deletion - Include guidance on CSI volumeHandle requirements when creating a new PV in the standby cluster that points to the same RBD image
Summary by CodeRabbit
✏️ Tip: You can customize this high-level summary in your review settings.