本文共 1032 字，大约阅读时间需要 3 分钟。

现象

2018-02-06T14:41:05.130+0800 I SHARDING [Balancer] moveChunk result: { cause: { ok: 0.0, errmsg: "can't accept new chunks because  there are still 1236 deletes from previous migration" }, ok: 0.0, errmsg: "moveChunk failed to engage TO-shard in the data transfer:  :: caused by :: UnknownError: can't accept new chunks because  there are still 1236 deletes...", code: 8 }

mongos集群数据分布不均匀，后台日志显示move chunk失败，我尝试手工迁移，报错一样

mongos> sh.moveChunk("recruit.externalResumeStorage", {"rid" : "1836776472", "sourceId" : 6}, "udb-f4rrmx")

{ "cause" : { "cause" : { "ok" : 0, "errmsg" : "can't accept new chunks because  there are still 1236 deletes from previous migration" }, "ok" : 0, "errmsg" : "moveChunk failed to engage TO-shard in the data transfer:  :: caused by :: UnknownError: can't accept new chunks because  there are still 1236 deletes from previous migration", "code" : 8 }, "code" : 8, "ok" : 0, "errmsg" : "move failed" }

故障原因

应该是bug，参考https://jira.mongodb.org/browse/SERVER-14047

解决方法

重启整个mongos集群，包括mongos节点，config节点，sharding节点

转载地址：http://ncbgb.baihongyu.com/