rai-setup: document the SQL-procedure reasoner surface, align engine sizes with platform docs - #41
Merged
Merged
Conversation
9 tasks
cafzal
force-pushed
the
setup-engine-management
branch
from
May 4, 2026 21:50
4600d94 to
1ce99cb
Compare
cafzal
changed the title
…sizes with platform docs The CLI and Python clients are thin wrappers over RELATIONALAI.API.* stored procedures. Surfacing the procedures directly closes a real gap: notebook/SQL-only workflows that don't run the CLI, and the case where the CLI version trails the backend on a new flag or size. Reasoner-agnostic — applies to logic + prescriptive equally. Predictive sizing/routing specifics ship with the predictive skills (PR #21). - references/engine-management.md: new "SQL stored procedures (canonical fallback)" section. Procedure table covers CREATE_REASONER / CREATE_REASONER_ASYNC, GET_REASONER, SUSPEND_REASONER, RESUME_REASONER_ASYNC, DELETE_REASONER, ALTER_REASONER_AUTO_SUSPEND_MINS, ALTER_REASONER_POOL_NODE_LIMITS, GET_JOB, CANCEL_JOB, plus the api.REASONERS and api.JOBS views. Aligned with the surface documented at docs.relational.ai/manage/compute-resources. Async + poll example. Generic "do not call RELATIONALAI.EXPERIMENTAL.*" callout (no reasoner-specific examples — those live with the relevant reasoner skills). - references/reasoners.md: Engine sizes table is now reasoner-aware (Logic + Prescriptive columns, AWS + Azure columns), reflecting the doc note that HIGHMEM_X64_L (AWS) and HIGHMEM_X64_SL (Azure) are Logic-only — Prescriptive does not currently accept the largest tier on either cloud. Adds a runtime note about the standard RELATIONAL_AI_<INSTANCE_FAMILY> compute-pool naming visible in GET_REASONER's RUNTIME field, with a link to the platform doc.
cafzal
force-pushed
the
setup-engine-management
branch
from
May 4, 2026 22:00
1ce99cb to
a1e4303
Compare
Contributor
|
This looks good. |
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.
Why
The CLI and Python clients (
rai reasoners:*,Resources,connect_sync()) are thin wrappers overRELATIONALAI.API.*stored procedures, butrai-setupdocuments only the wrappers — not the procedures themselves. That leaves a gap for notebook / SQL-only workflows, and for the case where the CLI version trails the backend on a new flag.Reasoner-agnostic — applies to logic and prescriptive engines. Predictive-specific sizing/routing (GPU sizes, ML-pool routing) and the
EXPERIMENTAL.*procedure ban (which only matters with a specific GNN-service example) ship with the predictive skills in #21.Changes
plugins/rai/skills/rai-setup/references/engine-management.md— new section "SQL stored procedures (canonical fallback)":CREATE_REASONER/CREATE_REASONER_ASYNC,GET_REASONER,SUSPEND_REASONER,RESUME_REASONER_ASYNC,DELETE_REASONER,ALTER_REASONER_AUTO_SUSPEND_MINS,ALTER_REASONER_POOL_NODE_LIMITS,GET_JOB,CANCEL_JOB, plus theapi.REASONERSandapi.JOBSviews.CREATE_REASONER_ASYNC→ loop onGET_REASONERuntilSTATUS=READY.plugins/rai/skills/rai-setup/references/reasoners.md:HIGHMEM_X64_L(AWS) andHIGHMEM_X64_SL(Azure) are Logic-only — Prescriptive does not currently accept the largest tier on either cloud.RELATIONAL_AI_<INSTANCE_FAMILY>compute-pool naming visible inGET_REASONER'sRUNTIMEfield.Test plan
rai reasoners:create/delete/suspend/resume/list/get/alter) match the documented CLI surface.HIGHMEM_X64_LAWS-Logic-only,HIGHMEM_X64_SLAzure-Logic-only).GPU_NV_S, no_MLpool naming, noEXPERIMENTAL.*callout, noEXPERIMENTSschema discussion) — those land with PR Predictive skills: two-skill workflow (modeling + training) #21 in the predictive skills.Related
rai-predictive-modeling§ Engine sizing covering CPU vs GPU heuristics, the GNN-specific allow-list gap, and the_ML-suffixed compute-pool routing for predictive workloads. TheEXPERIMENTAL.*procedure ban (with theCREATE_GNN_SERVICEexample) lives there inrai-health§ Predictive train jobs stuck QUEUED.