Mongodb Profiler Output
如下是System.profile中的一条记录
{
"op" : "query",
"ns" : "omnisocials_admin_dev.GcLog",
"query" : { "$orderby" : { "createTime" : -1 }, "$query" : {} },
"ntoreturn" : 1,
"ntoskip" : 0,
"nscanned" : 0,
"nscannedObjects" : 164404,
"scanAndOrder" : true,
"keyUpdates" : 0,
"writeConflicts" : 0,
"numYield" : 1356,
"locks" : { "Global" : { "acquireCount" : { "r" : NumberLong(1357) } }, "MMAPV1Journal" : { "acquireCount" : { "r" : NumberLong(1357) } }, "Database" : { "acquireCount" : { "r" : NumberLong(1357) } }, "Collection" : { "acquireCount" : { "R" : NumberLong(1357) } } },
"nreturned" : 1,
"responseLength" : 421,
"millis" : 2348,
"execStats" : { "stage" : "OR", "nReturned" : 1, "executionTimeMillisEstimate" : 1860, "works" : 164482, "advanced" : 1, "needTime" : 164407, "needFetch" : 74, "saveState" : 1356, "restoreState" : 1356, "isEOF" : 0, "invalidates" : 0, "dupsTested" : 1, "dupsDropped" : 0, "locsForgotten" : 0, "matchTested_0" : 0, "matchTested_1" : 0, "inputStages" : [ { "stage" : "SORT", "nReturned" : 1, "executionTimeMillisEstimate" : 1850, "works" : 164482, "advanced" : 1, "needTime" : 164406, "needFetch" : 74, "saveState" : 1356, "restoreState" : 1356, "isEOF" : 1, "invalidates" : 0, "sortPattern" : { "createTime" : -1 }, "memUsage" : 409, "memLimit" : 3.35544e+07, "limitAmount" : 1, "inputStage" : { "stage" : "COLLSCAN", "filter" : { "$and" : [] }, "nReturned" : 164404, "executionTimeMillisEstimate" : 1130, "works" : 164480, "advanced" : 164404, "needTime" : 1, "needFetch" : 74, "saveState" : 1356, "restoreState" : 1356, "isEOF" : 1, "invalidates" : 0, "direction" : "forward", "docsExamined" : 164404 } }, { "stage" : "SORT", "nReturned" : 0, "executionTimeMillisEstimate" : 0, "works" : 0, "advanced" : 0, "needTime" : 0, "needFetch" : 0, "saveState" : 1356, "restoreState" : 1356, "isEOF" : 0, "invalidates" : 0, "sortPattern" : { "createTime" : -1 }, "memUsage" : 0, "memLimit" : 3.35544e+07, "inputStage" : { "stage" : "COLLSCAN", "filter" : { "$and" : [] }, "nReturned" : 0, "executionTimeMillisEstimate" : 0, "works" : 0, "advanced" : 0, "needTime" : 0, "needFetch" : 0, "saveState" : 1356, "restoreState" : 1356, "isEOF" : 0, "invalidates" : 0, "direction" : "forward", "docsExamined" : 0 } } ] },
"ts" : ISODate("2016-01-18T10:19:55.840Z"),
"client" : "127.0.0.1",
"allUsers" : [],
"user" : "" }如何生成system profile collection?
MongoDB Database Profiling 是一个捕获数据库执行活动的系统，它可以帮助识别慢查询和操作。
设置 profiling level
- 级别
- 级别０， 禁用；
- 级别１， 启用；只记录慢操作
- 级别２， 启用；记录所有操作
- 级别设置
- db.setProfilingLevel(level, slowms)
- level：级别
- slowms：慢操作阈值
- slowms默认100ms，所以使用db.setProfilingLevel(1)就表示设置记录慢操作，阈值100ms
- db.setProfilingLevel(level, slowms)
查询 profiling level & status
>db.getProfilingLevel()
1
>db.getProfilingStatus()
{ "was" : 1, "slowms" : 20 }
>db.system.profile.find().sort({$natural:-1})
{ "op" : "query", "ns" : "omnisocials_admin_dev.GcLog", "query" : { "$orderby" : { "createTime" : -1 }, "$query" : { } }, "ntoreturn" : 1, "ntoskip" : 0, "nscanned" : 0, "nscannedObjects" : 172989, "scanAndOrder" : true, "keyUpdates" : 0, "writeConflicts" : 0, "numYield" : 1351, "locks" : { "Global" : { "acquireCount" : { "r" : NumberLong(1352) } }, "MMAPV1Journal" : { "acquireCount" : { "r" : NumberLong(1352) } }, "Database" : { "acquireCount" : { "r" : NumberLong(1352) } }, "Collection" : { "acquireCount" : { "R" : NumberLong(1352) } } }, "nreturned" : 1, "responseLength" : 421, "millis" : 636, "execStats" : { "stage" : "OR", "nReturned" : 1, "executionTimeMillisEstimate" : 630, "works" : 172993, "advanced" : 1, "needTime" : 172992, "needFetch" : 0, "saveState" : 1351, "restoreState" : 1351, "isEOF" : 0, "invalidates" : 0, "dupsTested" : 1, "dupsDropped" : 0, "locsForgotten" : 0, "matchTested_0" : 0, "matchTested_1" : 0, "inputStages" : [ { "stage" : "SORT", "nReturned" : 1, "executionTimeMillisEstimate" : 630, "works" : 172993, "advanced" : 1, "needTime" : 172991, "needFetch" : 0, "saveState" : 1351, "restoreState" : 1351, "isEOF" : 1, "invalidates" : 0, "sortPattern" : { "createTime" : -1 }, "memUsage" : 409, "memLimit" : 33554432, "limitAmount" : 1, "inputStage" : { "stage" : "COLLSCAN", "filter" : { "$and" : [ ] }, "nReturned" : 172989, "executionTimeMillisEstimate" : 110, "works" : 172991, "advanced" : 172989, "needTime" : 1, "needFetch" : 0, "saveState" : 1351, "restoreState" : 1351, "isEOF" : 1, "invalidates" : 0, "direction" : "forward", "docsExamined" : 172989 } }, { "stage" : "SORT", "nReturned" : 0, "executionTimeMillisEstimate" : 0, "works" : 0, "advanced" : 0, "needTime" : 0, "needFetch" : 0, "saveState" : 1351, "restoreState" : 1351, "isEOF" : 0, "invalidates" : 0, "sortPattern" : { "createTime" : -1 }, "memUsage" : 0, "memLimit" : 33554432, "inputStage" : { "stage" : "COLLSCAN", "filter" : { "$and" : [ ] }, "nReturned" : 0, "executionTimeMillisEstimate" : 0, "works" : 0, "advanced" : 0, "needTime" : 0, "needFetch" : 0, "saveState" : 1351, "restoreState" : 1351, "isEOF" : 0, "invalidates" : 0, "direction" : "forward", "docsExamined" : 0 } } ] }, "ts" : ISODate("2016-01-19T04:56:49.725Z"), "client" : "127.0.0.1", "allUsers" : [ ], "user" : "" }
关于system.profile
Mongodb system.profile 是一个capped collection，capped collections是性能出色且有着固定大小的集合，对于大小固定，我们可以想象其就像一个环形队列，当集合空间用完后，再插入的元素就会覆盖最初始的头部的元素！
一些字段参考
| 相关字段 | 版本 | 含义 |
|---|---|---|
| op | 2.6+ | 操作类型，包括【insert，query，update，remove，getmore，command】 |
| ns | 2.6+ | 集合全路径，例如：omnisocials_admin_dev.GcLog |
| query | 2.6+ | 该操作document,如大于50kb则转对象为字符串摘要，如何摘要大于50kb,则进行截取 |
| command | 2.6+ | command操作文档,如大于50kb则转对象为字符串摘要，如何摘要大于50kb,则进行截取 |
| updateobj | 2.6+ | 在进行update操作时传递的update文档 |
| cursorid | 2.6+ | query和getmore操作时的游标ID |
| ntoreturn | 2.6+ | 操作指定返回的文档数量，例如：limit(5) |
| ntoskip | 2.6+ | skip是多少 |
| nscanned | 2.6+ | 当前操作执行时扫描的索引数量 |
| nscannedObjects | 2.6+ | 当前操作执行时扫描的集合数 |
| moved | 3.0* | 更新操作移动一个或多个文档到分区其他位置时，此值显示true |
| nmoved | 3.0* | 分区上移动的文档数 |
| scanAndOrder | 2.6+ | 是否以index的排序去返回请求结果 |
| ndeleted | 2.6+ | 当期操作删除的文档数量 |
| ninserted | 2.6+ | 当期操作插入的文档数量 |
| nMatched | 2.6+ | 仅适用与update操作 |
| nModified | 2.6+ | 当期操作修改的文档数量 |
| upsert | 2.6+ | A boolean that indicates the update operation’s upsert option value. Only appears if upsert is true. |
| keyUpdates | 2.6+ | 当前操作的执行引发了多少个index的改变，一个index的改变会引发性能消耗 |
| writeConflicts | 3.0+ | 如果当前操作是写操作，那么writeConfilicts则表示写操作时遇到的冲突数量 |
| numYield | 2.6+ | 当前操作在读取数据时，为使其他操作的执行二放弃的次数 |
| lockStats | 2.6- | Mongodb2.6引入的锁的监控（R）global read lock，（W）global write lock，（r）global write lock，（r）database specific read lock｜ |
| locks | 3.0+ | lock types and lock modes:[(Global)global lock,(MMAPV1Journal)MMAPv1 storage engine specific lock to synchronize journal writes,(Database)database lock.,(Collection)collection lock.,(Metadata)metadata lock.,(oplog)lock on the oplog.][(R)Shared lock,(W)Exclusive lock,(r)Intent Shared lock,(w)Intent Exclusive lock] |
| acquireCount | 3.0+ | 当前操作获取锁的次数 |
| acquireWaitCount | 3.0+ | 当前操作获取锁时需等待的次数 |
| timeAcquiringMicros | 3.0+ | 当前操作等待锁的总时间 |
| deadlockCount | 3.0+ | 当前操作等待锁释放时遇到死锁的次数 |
| nreturned | 2.6+ | 当前操作实际返回的文档数量 |
| responseLength | 2.6+ | 当前操作返回的文档大小 |
| millis | 2.6+ | 当前操作最终耗时 |
| execStats | 3.0+ | 包含执行统计操作的文档 |
| ts | 2.6+ | 当期操作开始时的时间 |
| client | 2.6+ | 发起操作的机器IP地址 |
| allUsers | 2.6+ | 从session获取的已认证的用户组 |
| user | 2.6+ | 执行当期操作的认证用户 |
参考链接
- Mongodb2.6：https://docs.mongodb.org/v2.6/reference/database-profiler/
- Mongodb3.0：https://docs.mongodb.org/v3.0/reference/database-profiler/