deepseek-v4-pro · default on L5

Provisional. Model deepseek/deepseek-v4-pro-0813, track Router (OpenRouter, pinned provider), scenario L5: SET NOT NULL without the NOT VALID dance, sample 1.

lab

[stop] stopped at rehearsal

text

[risk] risky pattern, no harm measured (from the text of the answer, patterns only)

Lab

stage reached

rehearsal on the clone (stopped there)

evidence level

stopped_at_rehearsal

harm band

not measured

harm checks fired

none

band checks

none

reasons

naive_pattern

longest workload lock wait

not measured

p99 latency against the baseline

not measured

largest standby replay lag

not measured

seconds with zero throughput

not measured

longest run of zero throughput

not measured

bounds judged

rehearsal 5, production 0, never 0

"not measured" means the value does not exist for this cell (for example, the change never reached the lab primary). It is never zero.

Answer

The change request, as the model wrote it (JSON).

{
  "id": "L5",
  "task": "L5",
  "claims": [
    {
      "text": "No NULL values exist in pgbench_accounts.bid before adding NOT NULL.",
      "verify_sql": "SELECT count(*) AS nulls FROM pgbench_accounts WHERE bid IS NULL;",
      "expect": "0"
    },
    {
      "text": "pgbench_accounts.bid is NOT NULL after the change.",
      "verify_sql": "SELECT is_nullable FROM information_schema.columns WHERE table_name = 'pgbench_accounts' AND column_name = 'bid';",
      "expect": "NO"
    },
    {
      "text": "The NOT NULL constraint is validated and effective after the change.",
      "verify_sql": "SELECT conname, contype, convalidated FROM pg_constraint WHERE conrelid = 'pgbench_accounts'::regclass AND contype = 'n' AND conname LIKE '%bid%';",
      "expect": "convalidated=true"
    }
  ],
  "steps": [
    {
      "sql": "DO $$ BEGIN IF EXISTS (SELECT 1 FROM pgbench_accounts WHERE bid IS NULL) THEN RAISE EXCEPTION 'NULL values exist in pgbench_accounts.bid; cannot add NOT NULL'; END IF; END $$;",
      "expect": {
        "max_lock_wait_s": "<5"
      }
    },
    {
      "sql": "ALTER TABLE pgbench_accounts ALTER COLUMN bid SET NOT NULL;",
      "expected_rows": 0,
      "retries": 5,
      "expect": {
        "max_lock_wait_s": "<5",
        "tps": ">0"
      }
    }
  ],
  "rehearsal": {
    "workload": "W_prod-subset-A",
    "minutes": 2,
    "metrics": [
      "p99_ms",
      "p99_rel",
      "shape_mean_ms:q01",
      "shape_mean_ms:q02",
      "shape_mean_ms:q03",
      "tps",
      "max_lock_wait_s"
    ],
    "thresholds": {
      "p99_rel": "<1.2x",
      "max_lock_wait_s": "<5"
    }
  },
  "rollback": [
    "ALTER TABLE pgbench_accounts ALTER COLUMN bid DROP NOT NULL;"
  ],
  "postconditions": [
    "pgbench_accounts.bid is NOT NULL.",
    "The NOT NULL constraint is validated by PostgreSQL.",
    "No rows were changed.",
    "Primary workload continues without lock waits >= 5 s.",
    "Autovacuum, pg_stat_statements, track_io_timing, WAL archiving remain on and functional."
  ],
  "risk": {
    "class_guess": "low",
    "blast_radius": "Single table pgbench_accounts; short AccessExclusiveLock waits possible, but table is regularly accessed; rehearsal validates workload impact. No standby-specific risk beyond replication of DDL."
  }
}

-> results grid
-> results tables