Rebuild: coordinator (planner, NNTP pool, spool, prefetcher, leases, rule sets), node API and the edge:node worker
Test and deploy / test (push) Successful in 21s
Test and deploy / deploy (push) Skipped

- groups:add / ranges:plan / nodes:create / coordinator:run commands
- the node API: lease, the range's chunks as one framed stream from the
  spool, batches, heartbeat, finish, release (node guard, token per node)
- edge:node: needs COORDINATOR_URL and NODE_TOKEN only (placeholders for
  the rest), posts a batch every 10 chunks, gives its range back on SIGTERM
- releases are gathered per batch: 1.04 chunk rows per release at 10
  chunks per batch (1.35 when built per XOVER chunk)
- partitions are checked per batch instead of remembered

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Daan MeijerandClaude Opus 5.5 committed 2026-10-09 00:31:06 +02:00
1 parent 20910647fa
commit d5f168d8d2
36 files changed
+1193 -14

No files matched your search

+94
View File
@@ -0,0 +1,94 @@
import type { HttpContext } from '@adonisjs/core/http'
import { Readable } from 'node:stream'
import { setTimeout as sleep } from 'node:timers/promises'
import Leases, { LeaseLost } from '#services/leases'
import BatchStore, { StaleBatch } from '#services/batch_store'
import Spool from '#services/spool'
import { encodeFrame, fromWire, type WireBatch } from '#lib/edge/index'
/** How long the stream waits for a chunk the prefetcher hasn't spooled yet. */
const CHUNK_WAIT_MS = 15 * 60 * 1000
/**
* The edge workers' API (REBUILD.md, parts 1–2): a lease, the range's
* chunks as one stream, results per batch, and the end of the range.
*/
export default class NodeController {
private node(ctx: HttpContext) {
return Number(ctx.auth.use('node').getUserOrFail().id)
}
async lease(ctx: HttpContext) {
const lease = await Leases.lease(this.node(ctx))
if (!lease) return ctx.response.noContent()
return lease
}
/** The range's chunks from its resume point, waiting for those still being fetched. */
async stream(ctx: HttpContext) {
const nodeId = this.node(ctx)
const rangeId = Number(ctx.params.id)
const range = await this.leased(ctx, rangeId, nodeId)
if (!range) return
const spans = Spool.chunksOf({ start: Number(range.start_id), end: Number(range.end_id) }, Number(range.next_id))
async function* frames() {
for (const span of spans) {
const waitUntil = Date.now() + CHUNK_WAIT_MS
while (!(await Spool.has(rangeId, span))) {
if (Date.now() > waitUntil) throw new Error(`Chunk ${span.from}–${span.to} of range ${rangeId} wasn't spooled in time`)
await sleep(500)
}
yield encodeFrame({ ...span, data: await Spool.read(rangeId, span) })
// Streaming keeps the lease alive.
await Leases.heartbeat(rangeId, nodeId)
}
}
ctx.response.header('Content-Type', 'application/octet-stream')
ctx.response.stream(Readable.from(frames()))
}
async batch(ctx: HttpContext) {
const nodeId = this.node(ctx)
const rangeId = Number(ctx.params.id)
try {
return await BatchStore.store(rangeId, nodeId, fromWire(ctx.request.body() as WireBatch))
} catch (error) {
if (error instanceof StaleBatch) return ctx.response.conflict({ error: error.message })
throw error
}
}
async heartbeat(ctx: HttpContext) {
return this.lifecycle(ctx, (rangeId, nodeId) => Leases.heartbeat(rangeId, nodeId))
}
async finish(ctx: HttpContext) {
return this.lifecycle(ctx, (rangeId, nodeId) => Leases.finish(rangeId, nodeId))
}
async release(ctx: HttpContext) {
return this.lifecycle(ctx, (rangeId, nodeId) => Leases.release(rangeId, nodeId))
}
private async lifecycle(ctx: HttpContext, action: (rangeId: number, nodeId: number) => Promise<void>) {
try {
await action(Number(ctx.params.id), this.node(ctx))
return { ok: true }
} catch (error) {
if (error instanceof LeaseLost) return ctx.response.conflict({ error: error.message })
throw error
}
}
private async leased(ctx: HttpContext, rangeId: number, nodeId: number) {
try {
return await Leases.leased(rangeId, nodeId)
} catch (error) {
if (error instanceof LeaseLost) {
ctx.response.conflict({ error: error.message })
return null
}
throw error
}
}
}