# IPC: CheckpointDelayStart

> 等待后端解除对检查点启动的阻塞。
---

<div class="event-kicker">PostgreSQL 等待事件档案</div>
<div class="event-identity">
  <span>类别<strong>IPC</strong></span>
  <span>事件<strong><code>CheckpointDelayStart</code></strong></span>
  <span>版本<strong>PG 17-18</strong></span>
  <span>证据<strong>2 个源码位置</strong></span>
</div>

## 官方描述译文 {#official-description}

等待后端解除对检查点启动的阻塞。

| PG 13 | PG 14 | PG 15 | PG 16 | PG 17 | PG 18 |
| :---: | :---: | :---: | :---: | :---: | :---: |
| — | — | — | — | ✓ | ✓ |

## 触发机制 {#mechanism}

`WAIT_EVENT_CHECKPOINT_DELAY_START` 位于 `src/backend/access/transam/xlog.c:7211`，是经 grep 核验的报告路径。公共身份/资源映射位于 `src/backend/utils/activity/wait_event_names.txt:116`。探针覆盖的操作是：等待后端解除对检查点启动的阻塞。CheckpointDelayStart 路径已经到达进程协调点，会休眠到对端、worker、屏障、队列或复制阶段发出进展信号。

## 正常还是麻烦？ {#normal-vs-trouble}

- <span class="severity-normal">**正常:**</span> 只要所有参与者都在继续推进，短暂汇合就属正常。
- <span class="severity-watch">**需要调查:**</span> 同一阶段连续三次采样仍存在、预期对端缺失或被阻塞，或队列及其依赖者停止推进时，应当调查。

## 诊断 SQL {#diagnostic-sql}

```sql {title="正在等待 IPC/CheckpointDelayStart 的会话"}
SELECT pid, backend_type, usename, datname, application_name,
       state, now() - query_start AS query_age,
       now() - xact_start AS xact_age,
       wait_event_type, wait_event,
       pg_blocking_pids(pid) AS blocking_pids,
       left(query, 160) AS query
FROM pg_stat_activity
WHERE wait_event_type = 'IPC'
  AND wait_event = 'CheckpointDelayStart'
ORDER BY query_age DESC NULLS LAST;
```

```sql {title="当前 IPC 等待分布"}
SELECT wait_event, count(*) AS waiting_sessions,
       count(*) FILTER (WHERE state = 'active') AS active_waiters,
       max(now() - query_start) AS oldest_query
FROM pg_stat_activity
WHERE wait_event_type = 'IPC'
GROUP BY wait_event
ORDER BY waiting_sessions DESC, wait_event;
```

```sql {title="这些会话的锁与关系上下文"}
SELECT a.pid, l.locktype, l.mode, l.granted, l.fastpath,
       d.datname, n.nspname, c.relname,
       l.page, l.tuple, l.virtualxid, l.transactionid,
       l.classid, l.objid, l.objsubid
FROM pg_stat_activity AS a
LEFT JOIN pg_locks AS l ON l.pid = a.pid
LEFT JOIN pg_database AS d ON d.oid = l.database
LEFT JOIN pg_class AS c
  ON c.oid = l.relation
 AND l.database = (
       SELECT oid FROM pg_database WHERE datname = current_database()
     )
LEFT JOIN pg_namespace AS n ON n.oid = c.relnamespace
WHERE a.wait_event_type = 'IPC'
  AND a.wait_event = 'CheckpointDelayStart'
ORDER BY a.pid, l.granted, l.locktype, l.mode;
```

## 处置建议 {#response}

1. 确认事件所指的对端或阶段。
2. 检查该参与者的等待、错误与进展状态。
3. 修复停滞的参与者或上游依赖，不要把等待者当作根因。

## 源码证据 {#source}

- [PostgreSQL 18.6 · src/backend/access/transam/xlog.c:7211](https://github.com/postgres/postgres/blob/REL_18_6/src/backend/access/transam/xlog.c#L7211) — `WAIT_EVENT_CHECKPOINT_DELAY_START`
- [PostgreSQL 18.6 · src/backend/utils/activity/wait_event_names.txt:116](https://github.com/postgres/postgres/blob/REL_18_6/src/backend/utils/activity/wait_event_names.txt#L116) — `CHECKPOINT_DELAY_START`

## 关联控制项与信号 {#related}

- **GUC:** [[guc:checkpoint_timeout]](/zh/guc/checkpoint-timeout/) · [[guc:max_wal_size]](/zh/guc/max-wal-size/)
- **指标:** [[metric:waiting_sessions]](/zh/metric/waiting-sessions/) · [[metric:wait_event_share]](/zh/metric/wait-event-share/) · [[metric:checkpoint_time]](/zh/metric/checkpoint-time/)