Question

我在mongodb主服务器上看到很多错误消息喜欢

    2016-07-14T15:15:13.657+0000 I STORAGE  [conn1264360] passes = 5000 in CappedRecordStoreV1::allocRecord: ns: local.oplog.rs, lenToAlloc: 3332496, maxCappedDocs: 9223372036854775807, nrecords: 
85038795, datasize: 52320193000. Continuing to delete old records to make room.
2016-07-14T15:15:24.835+0000 I STORAGE  [conn1264360] passes = 5000 in CappedRecordStoreV1::allocRecord: ns: local.oplog.rs, lenToAlloc: 3572180, maxCappedDocs: 9223372036854775807, nrecords: 
85033228, datasize: 52320114660. Continuing to delete old records to make room.
2016-07-14T15:29:56.434+0000 I STORAGE  [conn1265954] passes = 5000 in CappedRecordStoreV1::allocRecord: ns: local.oplog.rs, lenToAlloc: 3332760, maxCappedDocs: 9223372036854775807, nrecords: 
85035131, datasize: 52320300492. Continuing to delete old records to make room.
2016-07-14T15:30:01.777+0000 I STORAGE  [conn1265954] passes = 5000 in CappedRecordStoreV1::allocRecord: ns: local.oplog.rs, lenToAlloc: 3572748, maxCappedDocs: 9223372036854775807, nrecords: 
85029980, datasize: 52320290816. Continuing to delete old records to make room.

导致大量同步磁盘IO并且在某些时候Linux OS Ubuntu 14.04 LTS刚刚决定停止进程并等待IO完成

    [12114362.224095] INFO: task mongod:15282 blocked for more than 120 seconds.
[12114362.255630] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.

这导致mongo replicaset选举和故障转移到另一个节点。问题是，在mongodb节点进入SECONDARY状态后，此行为继续

2016-07-14T16:32:55.894+0000 I STORAGE  [rsSync] passes = 5000 in CappedRecordStoreV1::allocRecord: ns: local.oplog.rs, lenToAlloc: 3334112, maxCappedDocs: 9223372036854775807, nrecords: 85033970, datasize: 52320484040. Continuing to delete old records to make room.
2016-07-14T16:47:11.358+0000 I STORAGE  [rsSync] passes = 5000 in CappedRecordStoreV1::allocRecord: ns: local.oplog.rs, lenToAlloc: 3334116, maxCappedDocs: 9223372036854775807, nrecords: 85029709, datasize: 52320442564. Continuing to delete old records to make room.
2016-07-14T16:47:15.346+0000 I STORAGE  [rsSync] passes = 5000 in CappedRecordStoreV1::allocRecord: ns: local.oplog.rs, lenToAlloc: 3574168, maxCappedDocs: 9223372036854775807, nrecords: 85024240, datasize: 52320158028. Continuing to delete old records to make room.

Mongo 3.0.7版，mmap存储引擎我对脏页使用以下sysctl设置

vm.dirty_background_ratio = 5
vm.dirty_background_bytes = 0
vm.dirty_ratio = 25
vm.dirty_bytes = 0

知道发生了什么事吗？

mongodb＆＃34;继续删除旧记录以腾出空间＆＃34;

0 个答案: