Skip to content

docs: V2 data engine feature docs (RDMA transport, QoS limits, shallow/deep copy) and settings reference updates#1383

Open
jleeh wants to merge 1 commit into
longhorn:masterfrom
linkpoolio:pr/docs/v2-feature-docs
Open

docs: V2 data engine feature docs (RDMA transport, QoS limits, shallow/deep copy) and settings reference updates#1383
jleeh wants to merge 1 commit into
longhorn:masterfrom
linkpoolio:pr/docs/v2-feature-docs

Conversation

@jleeh

@jleeh jleeh commented Jun 25, 2026

Copy link
Copy Markdown

Summary

This PR adds user-facing documentation for V2 data engine features and updates the settings reference with new V2 settings.

New feature pages

RDMA Transport Support (v2-data-engine/rdma-transport.md)

Documents the NVMe-oF RDMA transport support for the V2 data engine:

  • Auto-detection of RDMA hardware via node labels (no manual configuration required)
  • Mixed-cluster support (TCP + RDMA nodes coexist)
  • Per-replica transport reporting
  • Verification commands

Per-Engine QoS Limits (v2-data-engine/engine-qos.md)

Documents per-volume QoS limits via SPDK bdev_set_qos_limit:

  • How to set QoS via StorageClass, Longhorn API, and kubectl
  • QoS parameter reference table (IOPS, read/write bandwidth)
  • Live QoS updates without detaching the volume
  • How limits are enforced (raid bdev level, rebuild traffic bypass)
  • Use cases (noisy neighbor isolation, bandwidth throttling, gradual ramp-up)

Shallow Copy and Deep Copy (v2-data-engine/shallow-deep-copy.md)

Documents the SPDK copy primitives used for v2 replica rebuilds:

  • Shallow copy (allocated clusters only)
  • Range shallow copy (incremental — copy only mismatching clusters)
  • Deep copy (full data duplication through snapshot chain)
  • Pipeline depth configuration
  • How rebuild uses these primitives (snapshot-granularity reuse, range copy for partial rebuilds)

Updated pages

Settings Reference (references/settings.md)

Added 4 new V2 data engine settings:

  • Data Engine Replica Controller Loss Timeout (ctrlr_loss_timeout_sec) — timeout before a lost NVMe-oF controller is destructed. Supports -1 (immediate failover), 0 (reconnect indefinitely), or positive value (seconds).
  • Data Engine Replica Reconnect Delay (reconnect_delay_sec) — delay between reconnection attempts to prevent tight retry loops.
  • Data Engine Replica Fast I/O Fail Timeout (fast_io_fail_timeout_sec) — timeout after which I/O to a disconnected controller fails fast.
  • Data Engine Shallow Copy Pipeline Depth (shallow_copy_pipeline_depth) — pipeline depth for shallow copy during rebuilds (default 1 = sequential).

V2 Data Engine Index (v2-data-engine/_index.md)

Updated with links to the 3 new feature pages and a brief overview of transport options and efficient rebuilds.

Related

@jleeh jleeh requested a review from a team as a code owner June 25, 2026 12:08
@jleeh jleeh force-pushed the pr/docs/v2-feature-docs branch from 048fb75 to 5396300 Compare June 25, 2026 12:09
@netlify

netlify Bot commented Jun 25, 2026

Copy link
Copy Markdown

Deploy Preview for longhornio ready!

Name Link
🔨 Latest commit 048fb75
🔍 Latest deploy log https://app.netlify.com/projects/longhornio/deploys/6a3d1a23f5280c0008b54adb
😎 Deploy Preview https://deploy-preview-1383--longhornio.netlify.app
📱 Preview on mobile
Toggle QR Code...

QR Code

Use your smartphone camera to open QR code link.

To edit notification comments on pull requests, go to your Netlify project configuration.

@netlify

netlify Bot commented Jun 25, 2026

Copy link
Copy Markdown

Deploy Preview for longhornio ready!

Name Link
🔨 Latest commit 81b9e67
🔍 Latest deploy log https://app.netlify.com/projects/longhornio/deploys/6a3d1d08ec44ab0008550da4
😎 Deploy Preview https://deploy-preview-1383--longhornio.netlify.app
📱 Preview on mobile
Toggle QR Code...

QR Code

Use your smartphone camera to open QR code link.

To edit notification comments on pull requests, go to your Netlify project configuration.

New feature pages:
- RDMA Transport Support: auto-detected dual-listener TCP+RDMA, mixed clusters
- Per-Engine QoS Limits: IOPS/bandwidth caps via SPDK bdev_set_qos_limit with usage guide
- Shallow Copy and Deep Copy: efficient rebuild primitives, pipeline depth, range copy

Updated settings reference:
- Data Engine Replica Controller Loss Timeout (ctrlr_loss_timeout_sec)
- Data Engine Replica Reconnect Delay (reconnect_delay_sec)
- Data Engine Replica Fast I/O Fail Timeout (fast_io_fail_timeout_sec)
- Data Engine Shallow Copy Pipeline Depth (shallow_copy_pipeline_depth)

Updated v2-data-engine index page with links to new feature pages

Signed-off-by: Jonny <jonny@linkpool.io>
@jleeh jleeh force-pushed the pr/docs/v2-feature-docs branch from 5396300 to 81b9e67 Compare June 25, 2026 12:20
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant