MongoServerError: 事务对于快照太旧:读取时间戳早于存储引擎快照历史中的最旧时间戳
MongoServerError: TransactionTooOldForSnapshot: Read timestamp is older than the oldest timestamp in the storage engine's snapshot history
ID: mongodb/transaction-too-old-for-snapshot
版本兼容性
| 版本 | 状态 | 引入 | 弃用 | 备注 |
|---|---|---|---|---|
| mongodb-4.0 | active | — | — | — |
| mongodb-4.2 | active | — | — | — |
| mongodb-4.4 | active | — | — | — |
| mongodb-5.0 | active | — | — | — |
| mongodb-6.0 | active | — | — | — |
| mongodb-7.0 | active | — | — | — |
根因分析
多文档事务使用的读取时间戳已从 WiredTiger 存储引擎的快照历史中清除,通常是由于长时间运行的事务或快照保留不足。
English
A multi-document transaction is using a read timestamp that has been pruned from the WiredTiger storage engine's snapshot history, typically due to long-running transactions or insufficient snapshot retention.
官方文档
https://www.mongodb.com/docs/manual/core/transactions/解决方案
-
Reduce the duration of transactions by committing them faster. Break large transactions into smaller batches. Example: instead of one transaction updating 10,000 documents, use 100 transactions of 100 documents each.
-
Increase the WiredTiger snapshot retention period by adjusting the `storage.wiredTiger.engineConfig.journalCompressor` or setting `eviction_dirty_target` to a higher value (requires server restart). Example: `db.adminCommand({setParameter:1,wiredTigerEngineConfig:{eviction_dirty_target:20}})` -
If using secondary reads, ensure the secondary's oplog window is large enough to support the transaction's read timestamp. Monitor with `rs.printSlaveReplicationInfo()` and increase oplog size if needed.
无效尝试
常见但无效的做法:
-
20% 失败
This does not affect snapshot retention; the issue is the storage engine pruning old snapshots, not the transaction timing out.
-
10% 失败
Restarting resets the snapshot history but does not prevent pruning; the error will recur if the same pattern occurs.
-
15% 失败
Changing read concern may affect transaction isolation guarantees and does not address the root cause of long-running operations.