精选留言(4)

• 

  鲨无赦

  唐老师好，问2个问题：
  1、分片集群下的副本集只有2台机器，一次意外断电导致其中一台系统启动不了，这个副本集也就坏了，如果强制把它从分片集群理彻底删除，现在一直处理【"draining" : true】。
  
  2、用MongoShake从A副本集同步数据到B副本集报错【Size must be between 0 and 16793600(16MB) First element】，2个副本集配置和参数都一样，那报错的这个文档为什么在A副本集写入是成功？

  2020-01-08

  

  
• 

  walker

  另外两种全量备份方式（文件复制、文件快照）后面会讲到吗？

  2020-01-05

  

  
• 

  cheriston

  [root@server22 ~]# mongorestore --host 192.168.127.22:27017 --oplogReplay
  2020-01-03T17:31:36.817+0800 using default 'dump' directory
  2020-01-03T17:31:36.817+0800 preparing collections to restore from
  2020-01-03T17:31:36.818+0800 reading metadata for test.random from dump/test/random.metadata.json
  2020-01-03T17:31:36.819+0800 restoring to existing collection config.cache.chunks.test.random without dropping
  2020-01-03T17:31:36.819+0800 reading metadata for config.cache.chunks.test.random from dump/config/cache.chunks.test.random.metadata.json
  2020-01-03T17:31:36.819+0800 restoring config.cache.chunks.test.random from dump/config/cache.chunks.test.random.bson
  2020-01-03T17:31:36.819+0800 restoring to existing collection config.cache.databases without dropping
  2020-01-03T17:31:36.819+0800 reading metadata for config.cache.databases from dump/config/cache.databases.metadata.json
  2020-01-03T17:31:36.819+0800 restoring to existing collection config.cache.collections without dropping
  2020-01-03T17:31:36.819+0800 reading metadata for config.cache.collections from dump/config/cache.collections.metadata.json
  2020-01-03T17:31:36.820+0800 restoring config.cache.databases from dump/config/cache.databases.bson
  2020-01-03T17:31:36.821+0800 finished restoring test.random (0 documents, 0 failures)
  2020-01-03T17:31:36.821+0800 restoring config.cache.collections from dump/config/cache.collections.bson
  2020-01-03T17:31:36.821+0800 finished restoring config.cache.chunks.test.random (0 documents, 0 failures)
  2020-01-03T17:31:36.822+0800 terminating read on config.cache.databases
  2020-01-03T17:31:36.822+0800 finished restoring config.cache.databases (0 documents, 0 failures)
  2020-01-03T17:31:36.822+0800 terminating read on config.cache.collections
  panic: close of closed channel

  2020-01-03

   1

  
• 

  撑伞也是雨中人。

  老师 分片集群具体如何实施备份？ 停掉均衡器之后，各个节点通过定时任务备份（各个机器启动备份时间点是否能够绝对一致，如果备份时间点存在几秒的差别，是否有致命风险？）

  作者回复: 停掉均衡以后，备份可以保证每个节点各自是准确的。但是各个节点之间时间点很难保证在同一点上，所以每个节点恢复以后那个状态不是完全同步的。这个没有致命风险，但是确实有数据不一致的可能性。如果要做到完全一致，需要使用MongoDB 企业版 Ops Manager的分片备份功能。

  2020-01-02

  

  