Tencent Cloud MongoDB Enhances maxTimeMS Handling to Avoid Service Avalanche

Background: In Tencent Cloud MongoDB, client request timeouts can lead to endless retries, causing server‑side request backlog and potential service avalanche. Existing overload protection mechanisms cause business impact and are not always timely.

Problem: Native MongoDB versions below 5.0 lack proper maxTimeMS propagation for write commands in sharded clusters and have no server‑side default maxTimeMS configuration, leaving clusters vulnerable to request pile‑up.

Native maxTimeMS principle: MongoDB creates an OperationContext for each request, tracking start time, elapsed time, and deadline. The checkForInterrupt() method terminates requests when shutdown, killOp, or hasDeadlineExpired() (via maxTimeMS) occurs.

Limitations in MongoDB <5.0: In sharded clusters, mongos applies maxTimeMS to the OperationContext but does not transmit it to mongod for write commands (insert/update/delete), so mongod never times out. Additionally, there is no built‑in default maxTimeMS setting.

Tencent Cloud optimizations: 1) Enhanced mongos to compute remaining time and attach maxTimeMS to each sub‑request sent to mongod, ensuring timeout propagation. 2) Added support for a server‑side default maxTimeMS via the config.cmongo_settings collection; user‑specified values override the default, and the setting syncs across mongos/mongod nodes within 10 seconds.

Usage tips: Upgrade to the latest 4.0/4.2 patch, then configure the default by inserting/updating db.cmongo_settings with _id: "maxTimeMS" and a value in milliseconds (e.g., 1000). To disable, set value to 0.

Conclusion: These changes prevent ineffective server‑side resource usage, reduce retry‑induced avalanche risk, and improve user experience by providing transparent, configurable timeout handling.