@sqlane/core 0.0.2 → 0.1.0

package/src/store/mysql_worker.js

@@ -0,0 +1,250 @@
+// The worker side of the store on MySQL: probe, candidates, claim. Times are computed in SQL, in UTC.
+
+/** @typedef {import('../types.js').Executor} Executor */
+/** @typedef {import('./store.js').ProbeRow} ProbeRow */
+/** @typedef {import('./store.js').Candidate} Candidate */
+/** @typedef {import('./store.js').ClaimParams} ClaimParams */
+/** @typedef {import('./store.js').ClaimedJob} ClaimedJob */
+
+/** The due jobs of each keyed name that the FIFO of the mutex keys reads */
+export const MUTEX_WINDOW_SIZE = 500;
+
+const NOW = 'UTC_TIMESTAMP(3)';
+/**
+ * @param {string} column
+ * @returns {string} The epoch milliseconds of a time column.
+ */
+const epochMs = column => `TIMESTAMPDIFF(MICROSECOND, '1970-01-01', ${column}) DIV 1000`;
+const HELD = `(m.held_until IS NULL OR m.held_until >= ${NOW})`;
+
+/**
+ * @param {number} count
+ * @returns {string}
+ */
+function placeholders(count) {
+    return Array.from({ length: count }, () => '?').join(', ');
+}
+
+/**
+ * @param {unknown} json
+ * @returns {string[]}
+ */
+function parseKeys(json) {
+    return json === null || json === undefined ? [] : JSON.parse(String(json));
+}
+
+/**
+ * The SQL of the probe. `GROUP BY status, job_name`: the whole index prefix, else MySQL does not use a loose index
+ * scan.
+ * @param {string} p
+ * @returns {string}
+ */
+export function probeSql(p) {
+    return `
+        SELECT job_name, TIMESTAMPDIFF(MICROSECOND, ${NOW}, MIN(run_at)) DIV 1000 AS delay_ms
+          FROM ${p}jobs FORCE INDEX (${p}jobs_status_name_run_at)
+         WHERE status = 'pending'
+         GROUP BY status, job_name
+    `;
+}
+
+/**
+ * The next due time of each job name with pending jobs: one index dive per pending name (loose index scan).
+ * @param {Executor} executor
+ * @param {string} p
+ * @returns {Promise<ProbeRow[]>}
+ */
+export async function probe(executor, p) {
+    const rows = await executor.query(probeSql(p), []);
+
+    return rows.map(row => ({ name: String(row.job_name), delayMs: Number(row.delay_ms) }));
+}
+
+/**
+ * @param {string} p
+ * @param {string[]} names
+ * @param {(index: number) => string} limitSql
+ * @returns {string} The oldest due jobs of each name, one index range per name.
+ */
+function dueJobsSql(p, names, limitSql) {
+    return names.map((_, index) => `(
+        SELECT id, job_name, run_at, mutex_keys FROM ${p}jobs FORCE INDEX (${p}jobs_status_name_run_at)
+         WHERE status = 'pending' AND job_name = ? AND run_at <= ${NOW}
+         ORDER BY run_at, id LIMIT ${limitSql(index)}
+    )`).join(' UNION ALL ');
+}
+
+/**
+ * The due jobs that can start now, oldest first: at most `limit` of each name. A job of a keyed name waits if one of
+ * its keys is held, or an older due job of a keyed name has it (FIFO per key over the window of the keyed names).
+ * @param {Executor} executor
+ * @param {string} p
+ * @param {Map<string, number>} limitsByName
+ * @param {string[]} keyedNames The handled names whose definition has mutex keys.
+ * @returns {Promise<Candidate[]>}
+ */
+export async function findCandidates(executor, p, limitsByName, keyedNames) {
+    const keyed = new Set(keyedNames);
+    const plainNames = [...limitsByName.keys()].filter(name => !keyed.has(name));
+    const keyedRequested = [...limitsByName.keys()].filter(name => keyed.has(name));
+    /** @type {Record<string, unknown>[]} */
+    const rows = [];
+
+    if (plainNames.length) {
+        rows.push(...await executor.query(`
+            SELECT id, job_name, ${epochMs('run_at')} AS run_at_ms
+              FROM (${dueJobsSql(p, plainNames, () => '?')}) AS due
+        `, plainNames.flatMap(name => [name, limitsByName.get(name)])));
+    }
+
+    if (keyedRequested.length) {
+        const conditions = keyedRequested.map(() => '(job_name = ? AND name_rank <= ?)').join(' OR ');
+
+        rows.push(...await executor.query(`
+            WITH w AS (${dueJobsSql(p, keyedNames, () => String(MUTEX_WINDOW_SIZE))}),
+            wk AS (
+                SELECT w.id, k.mutex_key,
+                       ROW_NUMBER() OVER (PARTITION BY k.mutex_key ORDER BY w.run_at, w.id) AS key_rank
+                  FROM w, JSON_TABLE(w.mutex_keys, '$[*]' COLUMNS (
+                      mutex_key VARCHAR(255) CHARACTER SET ascii COLLATE ascii_bin PATH '$'
+                  )) AS k
+            ),
+            busy AS (
+                SELECT wk.mutex_key FROM wk JOIN ${p}mutexes m ON m.mutex_key = wk.mutex_key
+                 WHERE wk.key_rank = 1 AND ${HELD}
+            ),
+            blocked AS (
+                SELECT id FROM wk WHERE key_rank > 1
+                 UNION SELECT wk.id FROM wk JOIN busy ON busy.mutex_key = wk.mutex_key
+            ),
+            eligible AS (
+                SELECT w.id, w.job_name, w.run_at,
+                       ROW_NUMBER() OVER (PARTITION BY w.job_name ORDER BY w.run_at, w.id) AS name_rank
+                  FROM w WHERE NOT EXISTS (SELECT 1 FROM blocked b WHERE b.id = w.id)
+            )
+            SELECT id, job_name, ${epochMs('run_at')} AS run_at_ms FROM eligible WHERE ${conditions}
+        `, [...keyedNames, ...keyedRequested.flatMap(name => [name, limitsByName.get(name)])]));
+    }
+
+    return rows
+        .map(row => ({ id: String(row.id), name: String(row.job_name), runAtMs: Number(row.run_at_ms) }))
+        .toSorted((a, b) => a.runAtMs - b.runAtMs || (a.id < b.id ? -1 : 1));
+}
+
+/**
+ * Takes the keys of the picked jobs, all or none per job. A held key stays with its holder; an ended linger is taken
+ * over. Keys are inserted in ascending order: no deadlocks between claims.
+ * @param {Executor} executor
+ * @param {string} p
+ * @param {{ id: string, mutexKeys: readonly string[] }[]} jobs The picked keyed jobs: their keys are distinct.
+ * @param {string} lockToken
+ * @returns {Promise<Set<string>>} The ids of the jobs that got all their keys.
+ */
+async function takeKeys(executor, p, jobs, lockToken) {
+    const holds = jobs
+        .flatMap(job => job.mutexKeys.map(key => ({ key, id: job.id })))
+        .toSorted((a, b) => (a.key < b.key ? -1 : 1));
+    const keys = holds.map(hold => hold.key);
+    // held_until last: MySQL sets the columns from left to right, and the others read its old value
+    /**
+     * @param {string} column
+     * @returns {string} Takes the new value over an ended linger.
+     */
+    const takeOver = column => `${column} = IF(${p}mutexes.held_until < ${NOW}, new.${column}, ${p}mutexes.${column})`;
+
+    await executor.query(`
+        INSERT INTO ${p}mutexes (mutex_key, job_id, lock_token, held_until, created_at)
+        VALUES ${holds.map(() => `(?, ?, ?, NULL, ${NOW})`).join(', ')} AS new
+        ON DUPLICATE KEY UPDATE ${takeOver('job_id')}, ${takeOver('lock_token')}, ${takeOver('created_at')},
+            held_until = IF(${p}mutexes.held_until < ${NOW}, NULL, ${p}mutexes.held_until)
+    `, holds.flatMap(hold => [hold.key, hold.id, lockToken]));
+
+    const acquired = new Set((await executor.query(`
+        SELECT mutex_key FROM ${p}mutexes WHERE mutex_key IN (${placeholders(keys.length)}) AND lock_token = ?
+    `, [...keys, lockToken])).map(row => String(row.mutex_key)));
+    const withAllKeys = new Set(jobs.filter(job => job.mutexKeys.every(key => acquired.has(key))).map(job => job.id));
+    const partial = holds.filter(hold => !withAllKeys.has(hold.id) && acquired.has(hold.key));
+
+    if (partial.length) {
+        await executor.query(`
+            DELETE FROM ${p}mutexes WHERE mutex_key IN (${placeholders(partial.length)}) AND lock_token = ?
+        `, [...partial.map(hold => hold.key), lockToken]);
+    }
+
+    return withAllKeys;
+}
+
+/**
+ * Claims due jobs among the candidates. Call it in a queue transaction.
+ * @param {Executor} executor
+ * @param {string} p
+ * @param {ClaimParams} params
+ * @returns {Promise<ClaimedJob[]>} In the order of run_at.
+ */
+export async function claim(executor, p, { candidateIds, lockToken, lockedBy, leaseMs, pick }) {
+    if (!candidateIds.length) {
+        return [];
+    }
+
+    const lockedRows = (await executor.query(`
+        SELECT id, job_name, CAST(mutex_keys AS CHAR) AS mutex_keys, attempts, max_attempts, stalls, max_stalls,
+               failure_job, ${epochMs('created_at')} AS created_at_ms,
+               TIMESTAMPDIFF(MICROSECOND, run_at, ${NOW}) DIV 1000 AS waited_ms
+          FROM ${p}jobs
+         WHERE id IN (${placeholders(candidateIds.length)}) AND status = 'pending' AND run_at <= ${NOW}
+         ORDER BY run_at, id
+           FOR UPDATE SKIP LOCKED
+    `, candidateIds)).map(row => ({
+        id: String(row.id),
+        name: String(row.job_name),
+        mutexKeys: parseKeys(row.mutex_keys),
+        attempts: Number(row.attempts),
+        maxAttempts: Number(row.max_attempts),
+        stalls: Number(row.stalls),
+        maxStalls: Number(row.max_stalls),
+        failureJob: row.failure_job === null ? null : String(row.failure_job),
+        enqueuedAt: new Date(Number(row.created_at_ms)),
+        waitedMs: Number(row.waited_ms),
+        lockToken,
+    }));
+    const lockedKeys = [...new Set(lockedRows.flatMap(row => row.mutexKeys))];
+    const busyKeys = lockedKeys.length
+        ? new Set((await executor.query(`
+            SELECT m.mutex_key FROM ${p}mutexes m WHERE m.mutex_key IN (${placeholders(lockedKeys.length)}) AND ${HELD}
+        `, lockedKeys)).map(row => String(row.mutex_key)))
+        : new Set();
+    const picked = pick(lockedRows, busyKeys);
+    const keyedPicked = picked.filter(row => row.mutexKeys.length);
+    const withKeys = keyedPicked.length ? await takeKeys(executor, p, keyedPicked, lockToken) : new Set();
+    const claimed = picked.filter(row => !row.mutexKeys.length || withKeys.has(row.id));
+
+    if (claimed.length) {
+        await executor.query(`
+            UPDATE ${p}jobs
+               SET status = 'running', lock_token = ?, locked_by = ?,
+                   locked_until = ${NOW} + INTERVAL ? MICROSECOND, started_at = ${NOW}, timed_out_at = NULL,
+                   updated_at = ${NOW}
+             WHERE id IN (${placeholders(claimed.length)})
+        `, [lockToken, lockedBy, leaseMs * 1000, ...claimed.map(row => row.id)]);
+    }
+
+    return claimed;
+}
+
+/**
+ * @param {Executor} executor
+ * @param {string} p
+ * @param {string[]} ids
+ * @returns {Promise<Map<string, unknown>>} The payloads by job id, as JSON gives them back.
+ */
+export async function readPayloads(executor, p, ids) {
+    if (!ids.length) {
+        return new Map();
+    }
+
+    const rows = await executor.query(`
+        SELECT id, CAST(payload AS CHAR) AS payload FROM ${p}payloads WHERE id IN (${placeholders(ids.length)})
+    `, ids);
+
+    return new Map(rows.map(row => [String(row.id), JSON.parse(String(row.payload))]));
+}

package/src/store/postgres_enqueue.js

@@ -0,0 +1,54 @@
+// Enqueue on PostgreSQL: one statement, atomic even outside a transaction.
+
+/**
+ * Inserts the jobs and their payloads.
+ * @param {import('../types.js').Executor} executor
+ * @param {string} p The table prefix.
+ * @param {import('../new_job.js').NewJob[]} jobs Their ids are distinct.
+ * @param {string} enqueueToken
+ * @returns {Promise<Set<string>>} The ids that this call created.
+ */
+export async function insertJobsPostgres(executor, p, jobs, enqueueToken) {
+    const input = jobs.map(job => ({
+        id: job.id,
+        job_name: job.name,
+        payload: JSON.parse(job.payloadJson),
+        mutex_keys: job.mutexKeys.length ? job.mutexKeys : null,
+        max_attempts: job.maxAttempts,
+        max_stalls: job.maxStalls,
+        completed_retention_ms: job.completedRetentionMs,
+        failed_retention_ms: job.failedRetentionMs,
+        failure_job: job.failureJob,
+        delay_ms: job.delayMs,
+        run_at: job.runAt?.toISOString() ?? null,
+    }));
+    const rows = await executor.query(`
+        WITH input AS (
+            SELECT * FROM jsonb_to_recordset($1::jsonb) AS r(
+                  id text, job_name text, payload jsonb, mutex_keys text[], max_attempts integer,
+                  max_stalls integer, completed_retention_ms bigint, failed_retention_ms bigint, failure_job text,
+                  delay_ms bigint, run_at timestamptz
+            )
+        ),
+        inserted AS (
+            INSERT INTO ${p}jobs (id, job_name, status, run_at, max_attempts, max_stalls, mutex_keys, enqueue_token,
+                completed_retention_ms, failed_retention_ms, failure_job, created_at, updated_at)
+            SELECT id, job_name, 'pending',
+                   -- Truncated: timestamptz(3) rounds, and a rounded-up time is not due for the claim
+                   COALESCE(run_at, date_trunc('milliseconds', clock_timestamp() + delay_ms * interval '1 millisecond')),
+                   max_attempts, max_stalls, mutex_keys, $2, completed_retention_ms, failed_retention_ms, failure_job,
+                   clock_timestamp(), clock_timestamp()
+              FROM input ORDER BY id
+            ON CONFLICT (id) DO NOTHING
+            RETURNING id
+        ),
+        payloads AS (
+            INSERT INTO ${p}payloads (id, payload)
+            SELECT input.id, input.payload FROM input JOIN inserted ON inserted.id = input.id
+            ON CONFLICT (id) DO UPDATE SET payload = EXCLUDED.payload
+        )
+        SELECT id FROM inserted
+    `, [JSON.stringify(input), enqueueToken]);
+
+    return new Set(rows.map(row => String(row.id)));
+}