Doca Erasure Coding Agent Skill

DOCA Erasure Coding

Where to start: This skill assumes DOCA is already installed and

the user is doing hands-on erasure-coding work on a BlueField /

ConnectX / host with DOCA. Open TASKS.md if the user

wants to *do* something (configure / build / modify / run / test /

debug); open CAPABILITIES.md when the question

is *what can DOCA Erasure Coding express* on this version. If the

user has not installed DOCA yet, route to

doca-setup first. If the user is

asking *"should I even use erasure coding here, or just replicate

the data?"*, the path-selection rule in

CAPABILITIES.md ## Capabilities and modes

is the first stop — erasure coding is a storage-resilience

primitive (RAID-6 / distributed file system parity / object-storage

erasure-coded buckets), not a network primitive and not a

replication substitute.

Example questions this skill answers well

The CLASSES of DOCA Erasure Coding questions this skill is built

to answer, each with one worked example. The agent should treat

the *class* as the load-bearing piece — the worked example is a

single instance.

• **"Should I offload this erasure coding to DOCA, or do it on the

CPU — or just replicate the data instead?"** — worked example:

*"I am writing a distributed object store that wants RAID-6-like

resilience across 12 disks; is doca-erasure-coding the right

primitive, or should I just keep three copies?"*. Answered by

the path-selection table in

CAPABILITIES.md ## Capabilities and modes

+ the *"when NOT to use doca-erasure-coding"* bullets in

CAPABILITIES.md ## Safety policy.

• **"Does my device support the EC task type, block size, N+K

layout, and matrix type I want?"** — worked example: *"is

doca_ec_task_create on this BlueField, what is the maximum

block size, what is the cap on N+K, and does the device

advertise a Vandermonde Reed-Solomon matrix?"*. Answered by

the per-task + matrix + size capability-query rule

(doca_ec_cap_task_create_is_supported,

_task_recover_is_supported, _task_update_is_supported,

_task_galois_mul_is_supported — the 4th public task on the

EC accelerator, doca_ec_cap_get_max_block_size,

doca_ec_cap_get_max_buf_list_len, and per-variant

doca_ec_matrix_create() constructor success — the public

header does NOT ship a doca_ec_cap_get_matrix_* family) in

CAPABILITIES.md ## Capabilities and modes

+ the discovery step in

TASKS.md ## configure.

• **"One source block changed — do I have to recompute all the

parity from scratch?"** — worked example: *"I have 8 data + 4

parity blocks on disk; one data block was just rewritten. How

do I express the parity update without re-running the full

encode?"*. Answered by the doca_ec_task_update row in the

task-type table in

CAPABILITIES.md ## Capabilities and modes

+ the update-vs-re-encode rule in

CAPABILITIES.md ## Safety policy

+ the modify-from-sample slot for switching between create and

update tasks in

TASKS.md ## modify.

• **"K of my N+K blocks went missing — how do I reconstruct

them?"** — worked example: *"a disk failed; the 8 data blocks

are intact but 2 of the 4 parity blocks are gone; can DOCA

recover them?"*. Answered by the

doca_ec_task_recover row in the task-type table in

CAPABILITIES.md ## Capabilities and modes

+ the recover smoke (encode → drop a block → recover → bit-compare)

in TASKS.md ## test.

• "What permissions does the source / destination mmap need?" —

worked example: *"my doca_ec_task_create returns

DOCA_ERROR_NOT_PERMITTED"*. Answered by the permission matrix

in CAPABILITIES.md ## Safety policy

+ the mmap-set-permissions checklist in

TASKS.md ## test.

• **"Is this DOCA Erasure Coding API available on my installed

DOCA version?"** — worked example: *"is doca_ec_task_update in

the DOCA I have installed, or do I have to recompute parity from

scratch?"*. Answered by the version-compatibility overlay in

CAPABILITIES.md ## Version compatibility,

which cross-links the canonical detection chain in

doca-version and adds the

EC-specific *"discover per-task support + matrix type via cap

query"* bullets.

• **"What does this DOCA_ERROR_* from an EC call mean and which

layer caused it?"** — worked example: *"DOCA_ERROR_INVALID_VALUE

on doca_ec_task_create_allocate_init with block_size = 4 MiB"*.

Answered by the EC overlay on the cross-library taxonomy in

CAPABILITIES.md ## Error taxonomy

+ the layered ladder in

TASKS.md ## debug that escalates to

doca-debug.

Audience

This skill serves **external developers building applications that

consume the DOCA Erasure Coding library** — i.e., users whose code

calls doca_ec_* (directly in C/C++, or through FFI/bindings from

another language) to offload Reed-Solomon erasure coding onto a

BlueField DPU or ConnectX accelerator. It is *not* for NVIDIA

developers contributing to DOCA Erasure Coding itself.

The canonical fit is distributed storage: RAID-6-style block

layouts, distributed file system parity, object-storage

erasure-coded buckets, and any data-durability workload where the

N data + K redundancy block model lets the system tolerate K

simultaneous block losses without data loss. The skill keeps the

agent oriented to that domain: erasure coding is operationally

distinct from pure replication (which keeps M whole copies) and

from RAID-6 the disk layout (which is one specific instance of

the N=k, K=2 case); confusing them produces wrong recommendations.

Language scope. DOCA Erasure Coding ships as a C library with

pkg-config module name doca-erasure-coding. The shipped samples

are written in C. C and C++ consumers are the canonical case and

the worked examples in TASKS.md assume that path. Other-language

consumers (Rust, Go, Python, …) consume the same *.so through

FFI or language-specific bindings; the skill's contribution in

that case is to keep the lifecycle, capability-discovery,

permission, error-taxonomy, and create-vs-recover-vs-update

guidance language-neutral, and to route the agent to the public C

ABI as the authoritative surface that any wrapper will eventually

call.

When to load this skill

Load this skill when the user is doing hands-on DOCA Erasure

Coding work, in any language. Concretely:

• Initializing a doca_ec context on a doca_dev and configuring

at least one task type (doca_ec_task_create,

doca_ec_task_recover, and/or doca_ec_task_update) before

doca_ctx_start().

• Choosing between the create task (encode N data blocks into

K redundancy blocks), the recover task (reconstruct up to K

missing blocks given the surviving subset), and the update

task (incrementally update redundancy blocks when one source

block changes — much cheaper than re-encoding from scratch).

• Picking the coding parameters: matrix type / coding scheme

(typically Vandermonde or Cauchy for Reed-Solomon), N (number

of data blocks), K (number of redundancy blocks), and block

size (all blocks the same size).

• Setting permissions on doca_mmap correctly for the source

buffers — data blocks for create, data + parity for recover —

(DOCA_ACCESS_FLAG_LOCAL_READ_ONLY at minimum) and the

destination buffers — redundancy blocks for create, recovered

blocks for recover, updated parity for update —

(DOCA_ACCESS_FLAG_LOCAL_READ_WRITE).

• Sizing blocks against

doca_ec_cap_get_max_block_size(devinfo) and N+K against

doca_ec_cap_get_max_buf_list_len(devinfo, &max_buf_list_len).

• Checking which task types and matrix schemes this device's

accelerator advertises via the

doca_ec_cap_task_*_is_supported (including the 4th task

_task_galois_mul_is_supported) and per-variant

doca_ec_matrix_create() constructor success — the public

header does NOT ship a doca_ec_cap_get_matrix_*

families against the active doca_devinfo.

• Validating against a known-vector recover smoke (encode a small

N + K layout, simulate one block loss, recover, bit-compare

against the original block) before pushing bulk storage data

through the accelerator.

• Recognising when update is the right task (single source

block changed in an existing N + K layout — doca_ec_task_update

is far cheaper than recomputing all K parity blocks via a fresh

doca_ec_task_create).

• Debugging a DOCA_ERROR_* returned from an EC call (lifecycle

vs. block-size vs. N+K vs. matrix-type vs. permission vs.

queue pressure) and the task-completion event on the progress

engine.

• Designing or extending non-C bindings (Rust, Go, Python, …)

that wrap the EC C ABI — for the lifecycle, permission,

capability, and create-vs-recover-vs-update rules the wrapper

must honor.

Do not load this skill for general DOCA orientation, install

of DOCA itself, non-Reed-Solomon erasure codes (fountain codes,

LDPC — those belong on a CPU library, not the DOCA accelerator),

pure replication / mirroring designs (erasure coding is the wrong

primitive when one extra copy is the right answer), or other

DOCA libraries. For those, use

doca-public-knowledge-map.

What this skill provides

This is a thin loader. The body keeps only the orientation

needed to pick the right next file. The substantive

EC-specific material lives in two companion files:

• CAPABILITIES.md — what DOCA Erasure Coding can express on

this version: the three task types (create / recover / update,

each independently capability-gated), the matrix-type + N + K +

block-size configuration surface, the capability-query family

(doca_ec_cap_* for task support, max block size, max N+K,

supported matrix types), the EC error taxonomy (mapped onto

the cross-library DOCA_ERROR_* set), the observability

surface (per-task completion events on the progress engine),

the safety policy that gates source / destination mmap

permission decisions, and the path-selection rule (when to use

doca-erasure-coding versus a CPU EC library versus pure

replication versus RAID-6 on disk).

• TASKS.md — step-by-step workflows for the six in-scope EC

verbs: configure, build, modify, run, test, debug.

Plus a Deferred task verbs block that points out-of-scope

questions at the right next skill.

The skill assumes a host or BlueField where DOCA is already

installed at the standard location and the user has the

privileges their public install profile expects. It does not

cover installing DOCA — that path goes through

doca-setup.

What this skill deliberately does not ship

This skill is agent guidance, not a samples or templates

bundle. To keep the boundary clean, it deliberately does not

contain — and pull requests should not add:

• **Pre-written DOCA EC application source code, in any

language.** The verified EC source code is the shipped C

samples at /opt/mellanox/doca/samples/doca_erasure_coding/.

The agent's job is to route the user to those files and

prescribe a minimum-diff modification on them via the universal

modify-a-sample workflow in

doca-programming-guide,

layered with the EC-specific overrides in

TASKS.md ## modify.

• Pre-computed parity tables for arbitrary N + K layouts.

The skill tells the agent to use a *known-vector* smoke

(encode a tiny N + K layout from a fixed input, simulate

losing one block, recover, bit-compare against the original)

as the smoke before bulk; it does not ship a vector bank of

its own.

• Standalone build manifests (meson.build,

CMakeLists.txt, Cargo.toml, …) parked inside the skill.

The agent constructs the build manifest *in the user's

project directory* against the user's installed DOCA, where

pkg-config --modversion doca-erasure-coding is the source of

truth.

• A samples/, bindings/, or reference/ subtree of any

kind. A mock or incomplete artifact in this skill's tree, even

one labeled "reference", is misleading: users will read it as

buildable.

Loading order

• Read this SKILL.md first to confirm the user's question is in

scope (storage-domain resilience — N data + K redundancy

blocks — and not a network or replication question

misclassified as erasure coding).

• **For the EC capability matrix, the three task types

(create / recover / update), the matrix-type + N + K +

block-size configuration surface, the capability-query rules,

permission matrix, error taxonomy, observability, and safety /

path-selection policy, see CAPABILITIES.md.**

• **For step-by-step workflows — configure, build, modify, run,

test, debug — see TASKS.md.**

Both companion files cross-link to each other,

doca-version for the canonical

version-handling rules, and

doca-public-knowledge-map

whenever the right answer is "look it up in the public docs or

the installed package layout" rather than "EC-specific guidance".

Related skills

• doca-public-knowledge-map —

the routing table for every public DOCA documentation source

and the on-disk layout of an installed DOCA package. The DOCA

Erasure Coding page lives at

<https://docs.nvidia.com/doca/sdk/DOCA-Erasure-Coding/index.html>;

the install-tree layout (samples directory, header location)

belongs to the public knowledge map, not to this skill.

• doca-setup — env preparation,

install verification, and the *I have no install yet* path

with the public NGC DOCA container. This skill assumes its

preconditions are satisfied.

• doca-version — canonical DOCA

version-handling rules. This skill's

## Version compatibility

cross-links the four-way match rule and adds only the

EC-specific *"discover per-task support + supported matrix

types + max block size + max N+K via cap query"* overlay.

• doca-structured-tools-contract —

the bundle's structured-tools precedence rule (detect / prefer

/ fall back / report). The Command appendix in

TASKS.md honors this contract.

• doca-programming-guide —

general DOCA programming patterns shared by every library: the

canonical pkg-config + meson build pattern, the universal

modify-a-shipped-sample first-app workflow, the universal

lifecycle, the cross-library DOCA_ERROR_* taxonomy, and the

program-side debug order. This skill layers EC specifics on

top.

• doca-debug — the cross-cutting

debug ladder (install / version / build / link / runtime /

program / driver). EC-specific debug (task-not-supported,

matrix-type-unsupported, block-size-over-max, N+K-over-max,

recover-with-too-many-missing) overlays on top of that ladder.