deepseek-v4-pro · default on R8

Provisional. Model deepseek/deepseek-v4-pro-0813, track Router (OpenRouter, pinned provider), scenario R8: Raising a hot-standby-critical GUC on the primary first, sample 1.

lab

[inef] ineffective

text

[ok] right change (from the text of the answer, patterns only)

Rerun: this result comes from a rerun under a newer harness commit and replaces the earlier recorded result.

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

rehearsal_skipped_steps, rehearsal_failed

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 6, 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": "R8",
  "task": "R8",
  "claims": [
    {
      "text": "max_locks_per_transaction is currently 64 on the primary",
      "verify_sql": "select setting from pg_settings where name = 'max_locks_per_transaction'",
      "expect": "64"
    },
    {
      "text": "The standby is currently streaming and nearly caught up",
      "verify_sql": "select state, replay_lag from pg_stat_replication where application_name = 'walreceiver'",
      "expect": "streaming with replay_lag near zero"
    },
    {
      "text": "After the change, max_locks_per_transaction is 256 on both nodes",
      "verify_sql": "select setting from pg_settings where name = 'max_locks_per_transaction'",
      "expect": "256"
    },
    {
      "text": "Changing max_locks_per_transaction does not alter any table rows or schema objects",
      "verify_sql": "select count(*) from pg_class where relkind in ('r','p') and relnamespace = 'public'::regnamespace",
      "expect": "unchanged"
    }
  ],
  "steps": [
    {
      "config": {
        "name": "max_locks_per_transaction",
        "value": "256"
      },
      "node": "standby",
      "expect": {
        "setting:max_locks_per_transaction": "==256"
      }
    },
    {
      "config": {
        "name": "max_locks_per_transaction",
        "value": "256"
      },
      "node": "primary",
      "expect": {
        "setting:max_locks_per_transaction": "==256",
        "max_lock_wait_s": "<5"
      }
    },
    {
      "sql": "DO $$ BEGIN IF NOT EXISTS (SELECT 1 FROM pg_stat_replication WHERE application_name = 'walreceiver' AND state = 'streaming') THEN RAISE EXCEPTION 'standby is not streaming after restart'; END IF; END $$;",
      "expect": {
        "max_replication_lag_s": "<30"
      }
    }
  ],
  "rehearsal": {
    "workload": "W_prod-subset-B",
    "minutes": 2,
    "metrics": [
      "p99_ms",
      "p99_rel",
      "tps",
      "max_lock_wait_s",
      "io_util_pct",
      "server_version"
    ],
    "thresholds": {
      "p99_rel": "<1.2x",
      "max_lock_wait_s": "<5",
      "io_util_pct": "<90",
      "server_version": "==18.4"
    }
  },
  "rollback": [
    "ALTER SYSTEM RESET max_locks_per_transaction;",
    "SELECT 1;"
  ],
  "postconditions": [
    "max_locks_per_transaction is 256 on both the primary and the standby",
    "The standby is streaming from the primary",
    "Replication lag is below 30 seconds",
    "No table rows were modified",
    "pg_stat_statements, track_io_timing, autovacuum, and WAL archiving remain active"
  ],
  "risk": {
    "class_guess": "low-to-moderate: postmaster setting change requiring restarts on both nodes, but isolated to lock table sizing and approved for a maintenance window",
    "blast_radius": "both nodes restart briefly (excused by the task); existing sessions are dropped and reconnect; no data or schema changes"
  }
}

-> results grid
-> results tables