What causes "failed to write operator audit row" in thedotmack/claude-mem?

Check the logged `error` detail in the warn payload to identify the root SQL cause.

failed to write operator audit row

Error message

failed to write operator audit row

What it means

writeOperatorAudit inserts an audit_log row (via Postgres) recording CLI operator actions (retry/cancel) on observation-generation jobs. If the INSERT fails, it logs this warning and swallows the error — the operator action itself is not rolled back, only the audit trail entry is lost.

Solutions

Check the logged `error` detail in the warn payload to identify the root SQL cause.
Apply pending migrations so the audit_log table and its columns exist.
Ensure the DB role has INSERT permission on audit_log and pgcrypto/gen_random_uuid() is available.
Verify Postgres connectivity, then re-run the operator action if the audit row matters.

Example fix

// before: gen_random_uuid() undefined
CREATE EXTENSION IF NOT EXISTS pgcrypto;
// or on PG13+
-- nothing needed; verify table exists
\d audit_log
// after: retry the action
claude-mem server jobs retry <jobId>

Defensive patterns

Strategy: validation

Validate before calling

-- Before running operator actions, verify prerequisites
SELECT to_regclass('audit_log');            -- table exists?
SELECT gen_random_uuid();                   -- pgcrypto/PG13+ available?
SELECT has_table_privilege(current_user, 'audit_log', 'INSERT');

Type guard

function isAuditInsertError(e: unknown): boolean {
  const msg = e instanceof Error ? e.message : String(e);
  return /audit_log|permission denied|gen_random_uuid|relation .* does not exist/i.test(msg);
}

Try / catch

try {
  await pool.query(insertAuditSql, params);
} catch (error) {
  logger.warn('SYSTEM', 'failed to write operator audit row', {
    action,
    jobId: job.id,
    error: error instanceof Error ? error.message : String(error),
  }); // deliberate swallow: audit loss must not fail the operator action
}

Prevention

Apply all DB migrations before using jobs commands
Grant INSERT on audit_log to the CLI/server DB role
Enable pgcrypto or use PG13+ for gen_random_uuid()
Monitor Postgres connectivity in CI/operator environments

When it happens

Trigger: Running `jobs retry` or `jobs cancel` when the audit INSERT into audit_log fails — Postgres down, table missing, gen_random_uuid() unavailable (old PG without pgcrypto), FK violation on team_id/project_id/actor columns, or permission denied on audit_log.

Common situations: Database migration not applied (audit_log table absent); PostgreSQL < 13 without pgcrypto enabled; restricted DB role lacking INSERT on audit_log; transient connection loss.

Related errors

AI-assisted analysis of thedotmack/claude-mem@d8bc9755e7 (2026-09-17). Data as JSON: /api/errors/78897017564cbf38. Report an issue: GitHub.

Appendix: source

Thrown at src/npx-cli/commands/server-jobs.ts:430

interface PoolLike {
  query: (sql: string, params: unknown[]) => Promise<{ rows: unknown[] }>;
}

async function writeOperatorAudit(
  pool: { query: (sql: string, params: unknown[]) => Promise<unknown> },
  job: JobLookup,
  action: string,
  details: Record<string, unknown>,
): Promise<void> {
  try {
    await pool.query(
      `INSERT INTO audit_log (id, team_id, project_id, actor_id, api_key_id, action, resource_type, resource_id, details)
       VALUES (gen_random_uuid(), $1, $2, NULL, NULL, $3, 'observation_generation_job', $4, $5::jsonb)`,
      [job.team_id, job.project_id, action, job.id, JSON.stringify({ ...details, source: 'cli_operator' })],
    );
  } catch (error) {
    logger.warn('SYSTEM', 'failed to write operator audit row', {
      action,
      jobId: job.id,
      error: error instanceof Error ? error.message : String(error),
    });
  }
}

function extractRetriedCount(payload: Record<string, unknown> | null | undefined): number {
  if (!payload || typeof payload !== 'object') return 0;
  const value = (payload as { retried_count?: unknown }).retried_count;
  return typeof value === 'number' && Number.isFinite(value) && value >= 0 ? Math.floor(value) : 0;
}

function detectDivergence(
  pg: Record<string, number>,
  bullmq: Record<string, { waiting: number; active: number; completed: number; failed: number; delayed: number; stalled: number }> | null,
): Record<string, unknown> {
  if (!bullmq) return { reason: 'bullmq_unavailable' };

View on GitHub (pinned to d8bc9755e7)