Database Profiler Output
The database profiler captures data information about read and writeoperations, cursor operations, and database commands. To configure thedatabase profile and set the thresholds for capturing profile data,see the Database Profiler section.
The database profiler writes data in the system.profile
collection,which is a capped collection. To view the profiler’s output,use normal MongoDB queries on the system.profile
collection.
Note
Because the database profiler writes data to thesystem.profile
collection in adatabase, the profiler will profile some write activity, even fordatabases that are otherwise read-only.
Changed in version 3.4.
currentOp
and thedatabase profiler report the samebasic diagnostic information for all CRUD operations, including thefollowing:
aggregate
count
delete
distinct
find
(OP_QUERY andcommand
)findAndModify
getMore
(OP_GET_MORE andcommand
)insert
mapReduce
update
These operations are also included in the logging ofslow queries (see slowOpThresholdMs
formore information about slow query logging).
Example system.profile Document
The following presents some sample documents found in thesystem.profile
collection foroperations on a standalone:
- Find Operation
- Update Operation
The following document in the system.profile
reflects a find operation:
- {
- "op" : "query",
- "ns" : "test.report",
- "command" : {
- "find" : "report",
- "filter" : { "a" : { "$lte" : 500 } },
- "lsid" : {
- "id" : UUID("5ccd5b81-b023-41f3-8959-bf99ed696ce9")
- },
- "$db" : "test"
- },
- "cursorid" : 33629063128,
- "keysExamined" : 101,
- "docsExamined" : 101,
- "fromMultiPlanner" : true,
- "numYield" : 2,
- "nreturned" : 101,
- "queryHash" : "811451DD",
- "planCacheKey" : "759981BA",
- "locks" : {
- "Global" : {
- "acquireCount" : {
- "r" : NumberLong(3),
- "w" : NumberLong(3)
- }
- },
- "Database" : {
- "acquireCount" : { "r" : NumberLong(3) },
- "acquireWaitCount" : { "r" : NumberLong(1) },
- "timeAcquiringMicros" : { "r" : NumberLong(69130694) }
- },
- "Collection" : {
- "acquireCount" : { "r" : NumberLong(3) }
- }
- },
- "storage" : {
- "data" : {
- "bytesRead" : NumberLong(14736),
- "timeReadingMicros" : NumberLong(17)
- }
- },
- "responseLength" : 1305014,
- "protocol" : "op_msg",
- "millis" : 69132,
- "planSummary" : "IXSCAN { a: 1, _id: -1 }",
- "execStats" : {
- "stage" : "FETCH",
- "nReturned" : 101,
- "executionTimeMillisEstimate" : 0,
- "works" : 101,
- "advanced" : 101,
- "needTime" : 0,
- "needYield" : 0,
- "saveState" : 3,
- "restoreState" : 2,
- "isEOF" : 0,
- "docsExamined" : 101,
- "alreadyHasObj" : 0,
- "inputStage" : {
- ...
- }
- },
- "ts" : ISODate("2019-01-14T16:57:33.450Z"),
- "client" : "127.0.0.1",
- "appName" : "MongoDB Shell",
- "allUsers" : [
- {
- "user" : "someuser",
- "db" : "admin"
- }
- ],
- "user" : "someuser@admin"
- }
The profile entry for update
(anddelete
) operation contains the entire updatecommand.
The following example reflects an update
operation on a collection named report
.
- {
- "op" : "update",
- "ns" : "test.report",
- "command" : {
- "q" : { },
- "u" : { "$rename" : { "a" : "b" } },
- "multi" : true,
- "upsert" : false
- },
- "keysExamined" : 0,
- "docsExamined" : 25000,
- "nMatched" : 25000,
- "nModified" : 25000,
- "keysInserted" : 25000,
- "keysDeleted" : 25000000,
- "numYield" : 6985,
- "locks" : {
- "Global" : {
- "acquireCount" : { "r" : NumberLong(6986), "w" : NumberLong(13972) }
- },
- "Database" : {
- "acquireCount" : { "w" : NumberLong(6986) },
- "acquireWaitCount" : { "w" : NumberLong(1) },
- "timeAcquiringMicros" : { "w" : NumberLong(60899375) }
- },
- "Collection" : {
- "acquireCount" : { "w" : NumberLong(6986) }
- },
- "Mutex" : {
- "acquireCount" : { "r" : NumberLong(25000) }
- }
- },
- "storage" : {
- "data" : {
- "bytesRead" : NumberLong(126344060),
- "bytesWritten" : NumberLong(281834762),
- "timeReadingMicros" : NumberLong(94549),
- "timeWritingMicros" : NumberLong(139361)
- }
- },
- "millis" : 164687,
- "planSummary" : "COLLSCAN",
- "execStats" : {
- "stage" : "UPDATE",
- "nReturned" : 0,
- "executionTimeMillisEstimate" : 103771,
- "works" : 25003,
- "advanced" : 0,
- "needTime" : 25002,
- "needYield" : 0,
- "saveState" : 6985,
- "restoreState" : 6985,
- "isEOF" : 1,
- "nMatched" : 25000,
- "nWouldModify" : 25000,
- "wouldInsert" : false,
- "inputStage" : {
- "stage" : "COLLSCAN",
- "nReturned" : 25000,
- "executionTimeMillisEstimate" : 0,
- "works" : 25002,
- "advanced" : 25000,
- "needTime" : 1,
- "needYield" : 0,
- "saveState" : 31985,
- "restoreState" : 31985,
- "isEOF" : 1,
- "direction" : "forward",
- "docsExamined" : 25000
- }
- },
- "ts" : ISODate("2019-01-14T23:33:01.806Z"),
- "client" : "127.0.0.1",
- "appName" : "MongoDB Shell",
- "allUsers" : [
- {
- "user" : "someuser",
- "db" : "admin"
- }
- ],
- "user" : "someuser@admin"
- }
Output Reference
For any single operation, the documents created by the databaseprofiler will include a subset of the following fields. The preciseselection of fields in these documents depends on the type ofoperation.
Starting in MongoDB 4.2 (and in 4.0.9), for slow operations, the profilerentries anddiagnostic log messages includestorage
information.
Note
For the output specific to the version of your MongoDB, refer tothe appropriate version of the MongoDB Manual.
system.profile.
op
The type of operation. The possible values are:
command
count
distinct
geoNear
getMore
group
insert
mapReduce
query
remove
update
system.profile.
ns
- The namespace the operation targets. Namespaces in MongoDBtake the form of the database, followed by a dot (
.
),followed by the name ofthe collection.
Changed in version 3.6.
A document containing the full command object associated with thisoperation.
For example, the following output contains the command object for afind
operation on a collection named items
in adatabase named test
:
- "command" : {
- "find" : "items",
- "filter" : {
- "sku" : 1403978
- },
- ...
- "$db" : "test"
- }
The following example output contains the command object for agetMore
operation generated bya command with cursor id 19234103609
on a collection nameditems
in a database named test
:
- "command" : {
- "getMore" : NumberLong("19234103609"),
- "collection" : "items",
- "batchSize" : 10,
- ...
- "$db" : "test"
- },
If the command document exceeds 50 kilobytes, thedocument has the following form:
- "command" : {
- "$truncated": <string>,
- "comment": <string>
- }
The $truncated
field contains a string summary of the document excludingthe document’s comment
field if present. If the summary still exceeds50 kilobytes then it is further truncated, denoted by an ellipsis(…) at the end of the string.
The comment
field is present if a comment was passed to the operation.
Changed in version 3.6.
For "getmore"
operations which retrieve the next batch ofresults from a cursor, the originatingCommand
field contains thefull command object (e.g. find
or aggregate
) which originallycreated that cursor.
Changed in version 3.2.0.
Renamed from system.profile.nscanned
.The number of index keys that MongoDB scanned inorder to carry out the operation.
In general, if keysExamined
is much higherthan nreturned
, the database is scanning manyindex keys to find the result documents. Consider creating oradjusting indexes to improve query performance..
Changed in version 3.4.
keysExamined
is available for the following commands andoperations:
aggregate
find
(OP_QUERY andcommand
)findAndModify
count
distinct
getMore
(OP_GET_MORE andcommand
)mapReduce
delete
update
Changed in version 3.2.0: Renamed from system.profile.nscannedObjects
.
The number of documents in the collection that MongoDB scanned inorder to carry out the operation.
Changed in version 3.4.
docsExamined
is available for the following commands andoperations:
aggregate
find
(OP_QUERY andcommand
)findAndModify
count
distinct
getMore
(OP_GET_MORE andcommand
)mapReduce
delete
update
Changed in version 3.2.0: Renamed from system.profile.scanAndOrder
.
hasSortStage
is a boolean that is true
when a query cannot use the ordering in the index to return therequested sorted results; i.e. MongoDB must sort the documents afterit receives the documents from a cursor. The field only appears whenthe value is true
.
Changed in version 3.4.
hasSortStage
is available for the following commands andoperations:
find
(OP_QUERY andcommand
)getMore
(OP_GET_MORE andcommand
)findAndModify
mapReduce
aggregate
New in version 4.2.
A boolean that indicates whether any aggregation stage wrote data totemporary files due to memory restrictions.
Only appears if usedDisk
is true.
New in version 2.6.
The number of documents that match the system.profile.query
condition for the update operation.
New in version 2.6.
The number of documents modified by the update operation.
system.profile.
upsert
- A boolean that indicates the update operation’s
upsert
optionvalue. Only appears ifupsert
is true.
New in version 3.2.5.
A boolean that indicates whether the query planner evaluated multipleplans before choosing the winning execution plan for the query.
Only present when value is true
.
New in version 3.2.5.
A boolean that indicates whether the query system evicted a cachedplan and re-evaluated all candidate plans.
Only present when value is true
.
The number of index keys the update changed inthe operation. Changing an index key carries a small performancecost because the database must remove the old key and inserts a newkey into the B-tree index.
New in version 3.0.0.
The number of conflicts encountered during the write operation; e.g.an update
operation attempts to modify the same document asanother update
operation. See also write conflict.
system.profile.
numYield
- The number of times the operation yielded to allow other operationsto complete. Typically, operations yield when they need access todata that MongoDB has not yet fully read into memory. This allowsother operations that have data in memory to complete while MongoDBreads in data for the yielding operation. For more information,see the FAQ on when operations yield.
system.profile.
queryHash
- A hexadecimal string that represents a hash of the queryshape and is dependent only on the query shape.
queryHash
canhelp identify slow queries (including the query filter of writeoperations) with the same query shape.
Note
As with any hash function, two different query shapes may resultin the same hash value. However, the occurrence of hashcollisions between different query shapes is unlikely.
For more information on queryHash
and planCacheKey
, seequeryHash and planCacheKey.
New in version 4.2.
Unlike the queryHash
, theplanCacheKey
is a function ofboth the query shape and the currently available indexes forthat shape. That is, if indexes that can support the queryshape are added/dropped, the planCacheKey
value may changewhereas the queryHash
value would not change.
For more information on queryHash
and planCacheKey
, seequeryHash and planCacheKey.
New in version 4.2.
New in version 3.0.0: locks
replaces the lockStats
field.
The system.profile.locks
provides information for variouslock types and lock modes heldduring the operation.
The possible lock types are:
Lock TypeDescriptionParallelBatchWriterMode
Represents a lock for parallel batch writer mode.
In earlier versions, PBWM information was reported as part ofthe Global
lock information.
New in version 4.2.
ReplicationStateTransition
Represents lock taken for replica set member state transitions.
New in version 4.2.
Global
Represents global lock.Database
Represents database lock.Collection
Represents collection lock.Mutex
Represents mutex.Metadata
Represents metadata lock.oplog
Represents lock on the oplog.
The possible locking modes for the lock types are as follows:
Lock ModeDescriptionR
Represents Shared (S) lock.W
Represents Exclusive (X) lock.r
Represents Intent Shared (IS) lock.w
Represents Intent Exclusive (IX) lock.
The returned lock information for the various lock types include:
system.profile.locks.
acquireCount
Number of times the operation acquired the lock in the specifiedmode.
Number of times the operation had to wait for the
acquireCount
lock acquisitionsbecause the locks were held in a conflicting mode.acquireWaitCount
is less than orequal toacquireCount
.- Cumulative time in microseconds that the operation had to wait toacquire the locks.
timeAcquiringMicros
divided byacquireWaitCount
gives anapproximate average wait time for the particular lock mode.
system.profile.locks.
deadlockCount
- Number of times the operation encountered deadlocks while waitingfor lock acquisitions.
For more information on lock modes, seeWhat type of locking does MongoDB use?.
New in version 4.2: (Also available starting in 4.0.9)
The system.profile.storage
information provides metrics onthe storage engine data and wait time for the operation.
Specific storage metrics are returned only if the values are greaterthan zero.
New in version 4.2: (Also available starting in 4.0.9)
Number of bytes read by the operation from the disk to the cache.
New in version 4.2: (Also available starting in 4.0.9)
Time in microseconds that the operation had to spend to read fromthe disk.
New in version 4.2: (Also available starting in 4.0.9)
Number of bytes written by the operation from the cache to thedisk.
New in version 4.2: (Also available starting in 4.0.9)
Time in microseconds that the operation had to spend to write tothe disk.
New in version 4.2: (Also available starting in 4.0.9)
Time in microseconds that the operation had to wait for space inthe cache.
New in version 4.2: (Also available starting in 4.0.9)
Time in microseconds that the operation (if modifying the schema)had to wait to acquire a schema lock.
New in version 4.2: (Also available starting in 4.0.9)
Time in microseconds that the operation had to wait to acquirethe a lock on the needed data handles.
system.profile.
responseLength
The length in bytes of the operation’s result document. A large
responseLength
can affect performance.To limit the size of theresult document for a query operation, you can use any of thefollowing:
Note
When MongoDB writes query profile information to the log,the responseLength
value is in a fieldnamed reslen
.
system.profile.
protocol
- The MongoDB Wire Protocol request message format.
system.profile.
millis
- The time in milliseconds from the perspective of the
mongod
from the beginning of the operation to the end ofthe operation.
New in version 3.4.
A summary of the execution plan.
Changed in version 3.0.
A document that contains the execution statistics of the queryoperation. For other operations, the value is an empty document.
The system.profile.execStats
presents the statistics as atree; each node provides the statistics for the operation executedduring that stage of the query operation.
Note
The following fields list for execStats
is not meant to be exhaustive as the returned fields vary perstage.
New in version 3.0: stage
replaces the type
field.
The descriptive name for the operation performed as part of thequery execution; e.g.
- <code>COLLSCAN</code> for a collection scan
- <code>IXSCAN</code> for scanning index keys
- <code>FETCH</code> for retrieving documents
New in version 3.0: inputStages
replaces thechildren
field.
An array that contains statistics for the operations that are theinput stages of the current stage.
system.profile.
client
- The IP address or hostname of the client connection where theoperation originates.
New in version 3.4.
The identifier of the client application which ran the operation. Usethe appName
connection string option to set a custom valuefor the appName
field.
system.profile.
allUsers
- An array of authenticated user information (user name and database)for the session. See also Users.