mongodb runtime_error ai_generated partial

MongoServerError: balancer failed to move chunk: chunk { ns: "test.users", min: { _id: 1 }, max: { _id: 100 } } could not be moved

ID: mongodb/balancer-failed-to-move-chunk

Also available as: JSON · Markdown · 中文
78%Fix Rate
85%Confidence
1Evidence
2024-02-10First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
mongodb-5.0 active — — —
mongodb-6.0 active — — —
mongodb-7.0 active — — —

Root Cause

The balancer could not move a chunk because the target shard had insufficient storage or the chunk exceeded the maximum chunk size.

generic

中文

均衡器无法移动块，因为目标分片存储不足或块大小超过最大块大小。

Official Documentation

https://www.mongodb.com/docs/manual/core/sharding-balancer-administration/

Workarounds

  1. 80% success Check target shard disk usage with db.adminCommand({ listShards: 1 }) and add more shards or free up space on the target. 
    Check target shard disk usage with db.adminCommand({ listShards: 1 }) and add more shards or free up space on the target.
  2. 75% success Reduce chunk size temporarily via mongos config: use sh.enableAutoSplit() and sh.setBalancerState(true) with smaller chunkSize (e.g., 64MB) to split large chunks before migration. 
    Reduce chunk size temporarily via mongos config: use sh.enableAutoSplit() and sh.setBalancerState(true) with smaller chunkSize (e.g., 64MB) to split large chunks before migration.
  3. 70% success Manually move the chunk using sh.moveChunk() with a specific shard key range to force migration during low traffic. 
    Manually move the chunk using sh.moveChunk() with a specific shard key range to force migration during low traffic.

中文步骤

  1. Check target shard disk usage with db.adminCommand({ listShards: 1 }) and add more shards or free up space on the target.
  2. Reduce chunk size temporarily via mongos config: use sh.enableAutoSplit() and sh.setBalancerState(true) with smaller chunkSize (e.g., 64MB) to split large chunks before migration.
  3. Manually move the chunk using sh.moveChunk() with a specific shard key range to force migration during low traffic.

Dead Ends

Common approaches that don't work:

  1. 50% fail

    Increasing chunkSize to a very large value can cause uneven data distribution and balancer inefficiency.

  2. 50% fail

    Disabling the balancer entirely stops automatic data redistribution, leading to uneven shard loads.

  3. 50% fail

    Manually splitting chunks without considering data distribution can create many small chunks that strain the balancer.