0
0
mirror of https://github.com/mongodb/mongo.git synced 2024-12-01 09:32:32 +01:00
mongodb/debian/mongoperf.1
2015-01-30 18:18:58 -05:00

284 lines
5.7 KiB
Groff

.\" Man page generated from reStructuredText.
.
.TH "MONGOPERF" "1" "January 30, 2015" "3.0" "mongodb-manual"
.SH NAME
mongoperf \- MongoDB Performance Utility
.
.nr rst2man-indent-level 0
.
.de1 rstReportMargin
\\$1 \\n[an-margin]
level \\n[rst2man-indent-level]
level margin: \\n[rst2man-indent\\n[rst2man-indent-level]]
-
\\n[rst2man-indent0]
\\n[rst2man-indent1]
\\n[rst2man-indent2]
..
.de1 INDENT
.\" .rstReportMargin pre:
. RS \\$1
. nr rst2man-indent\\n[rst2man-indent-level] \\n[an-margin]
. nr rst2man-indent-level +1
.\" .rstReportMargin post:
..
.de UNINDENT
. RE
.\" indent \\n[an-margin]
.\" old: \\n[rst2man-indent\\n[rst2man-indent-level]]
.nr rst2man-indent-level -1
.\" new: \\n[rst2man-indent\\n[rst2man-indent-level]]
.in \\n[rst2man-indent\\n[rst2man-indent-level]]u
..
.SH SYNOPSIS
.sp
\fBmongoperf\fP is a utility to check disk I/O performance
independently of MongoDB.
.sp
It times tests of random disk I/O and presents the results. You can
use \fBmongoperf\fP for any case apart from MongoDB. The
\fBmmf\fP \fBtrue\fP mode is completely generic. In
that mode it is somewhat analogous to tools such as \fI\%bonnie++\fP (albeit mongoperf is
simpler).
.sp
Specify options to \fBmongoperf\fP using a JavaScript document.
.sp
\fBSEE ALSO:\fP
.INDENT 0.0
.INDENT 3.5
.INDENT 0.0
.IP \(bu 2
\fI\%bonnie\fP
.IP \(bu 2
\fI\%bonnie++\fP
.IP \(bu 2
\fI\%Output from an example run\fP
.IP \(bu 2
\fI\%Checking Disk Performance with the mongoperf Utility\fP
.UNINDENT
.UNINDENT
.UNINDENT
.SH OPTIONS
.INDENT 0.0
.TP
.B mongoperf
.UNINDENT
.INDENT 0.0
.TP
.B mongoperf
.UNINDENT
.INDENT 0.0
.TP
.B \-\-help, \-h
Returns information on the options and use of \fBmongoperf\fP\&.
.UNINDENT
.INDENT 0.0
.TP
.B <jsonconfig>
\fBmongoperf\fP accepts configuration options in the form of a
file that holds a \fIJSON\fP document. You must stream the
content of this file into \fBmongoperf\fP, as in the following
operation:
.INDENT 7.0
.INDENT 3.5
.sp
.nf
.ft C
mongoperf < config
.ft P
.fi
.UNINDENT
.UNINDENT
.sp
In this example \fBconfig\fP is the name of a file that holds a JSON
document that resembles the following example:
.INDENT 7.0
.INDENT 3.5
.sp
.nf
.ft C
{
nThreads:<n>,
fileSizeMB:<n>,
sleepMicros:<n>,
mmf:<bool>,
r:<bool>,
w:<bool>,
recSizeKB:<n>,
syncDelay:<n>
}
.ft P
.fi
.UNINDENT
.UNINDENT
.sp
See the \fI\%Configuration Fields\fP section for documentation of each
of these fields.
.UNINDENT
.SH CONFIGURATION FIELDS
.INDENT 0.0
.TP
.B mongoperf.nThreads
\fIType:\fP Integer.
.sp
\fIDefault:\fP 1
.sp
Defines the number of threads \fBmongoperf\fP will use in the
test. To saturate your system\(aqs storage system you will need
multiple threads. Consider setting \fBnThreads\fP to \fB16\fP\&.
.UNINDENT
.INDENT 0.0
.TP
.B mongoperf.fileSizeMB
\fIType:\fP Integer.
.sp
\fIDefault:\fP 1 megabyte (i.e. 1024\s-2\u2\d\s0 bytes)
.sp
Test file size.
.UNINDENT
.INDENT 0.0
.TP
.B mongoperf.sleepMicros
\fIType:\fP Integer.
.sp
\fIDefault:\fP 0
.sp
\fBmongoperf\fP will pause for the number of specified
\fBsleepMicros\fP divided by the
\fBnThreads\fP between each operation.
.UNINDENT
.INDENT 0.0
.TP
.B mongoperf.mmf
\fIType:\fP Boolean.
.sp
\fIDefault:\fP \fBfalse\fP
.sp
Set \fBmmf\fP to \fBtrue\fP to use memory mapped
files for the tests.
.sp
Generally:
.INDENT 7.0
.IP \(bu 2
when \fBmmf\fP is \fBfalse\fP, \fBmongoperf\fP
tests direct, physical, I/O, without caching. Use a large file
size to test heavy random I/O load and to avoid I/O coalescing.
.IP \(bu 2
when \fBmmf\fP is \fBtrue\fP, \fBmongoperf\fP
runs tests of the caching system, and can use normal file system
cache. Use \fBmmf\fP in this mode to test file system cache
behavior with memory mapped files.
.UNINDENT
.UNINDENT
.INDENT 0.0
.TP
.B mongoperf.r
\fIType:\fP Boolean.
.sp
\fIDefault:\fP \fBfalse\fP
.sp
Set \fBr\fP to \fBtrue\fP to perform reads as part of
the tests.
.sp
Either \fBr\fP or \fBw\fP must be \fBtrue\fP\&.
.UNINDENT
.INDENT 0.0
.TP
.B mongoperf.w
\fIType:\fP Boolean.
.sp
\fIDefault:\fP \fBfalse\fP
.sp
Set \fBw\fP to \fBtrue\fP to perform writes as part of
the tests.
.sp
Either \fBr\fP or \fBw\fP must be \fBtrue\fP\&.
.UNINDENT
.INDENT 0.0
.TP
.B mongoperf.recSizeKB
New in version 2.4.
.sp
\fIType:\fP Integer.
.sp
\fIDefault:\fP 4 kb
.sp
The size of each write operation.
.UNINDENT
.INDENT 0.0
.TP
.B mongoperf.syncDelay
\fIType:\fP Integer.
.sp
\fIDefault:\fP 0
.sp
Seconds between disk flushes. \fBmongoperf.syncDelay\fP is
similar to \fI\-\-syncdelay\fP for \fBmongod\fP\&.
.sp
The \fBsyncDelay\fP controls how frequently
\fBmongoperf\fP performs an asynchronous disk flush of the memory
mapped file used for testing. By default, \fBmongod\fP
performs this operation every 60 seconds. Use
\fBsyncDelay\fP to test basic system performance of
this type of operation.
.sp
Only use \fBsyncDelay\fP in conjunction with
\fBmmf\fP set to \fBtrue\fP\&.
.sp
The default value of \fB0\fP disables this.
.UNINDENT
.SH USE
.INDENT 0.0
.INDENT 3.5
.sp
.nf
.ft C
mongoperf < jsonconfigfile
.ft P
.fi
.UNINDENT
.UNINDENT
.sp
Replace \fBjsonconfigfile\fP with the path to the \fBmongoperf\fP
configuration. You may also invoke \fBmongoperf\fP in the
following form:
.INDENT 0.0
.INDENT 3.5
.sp
.nf
.ft C
echo "{nThreads:16,fileSizeMB:10000,r:true,w:true}" | mongoperf
.ft P
.fi
.UNINDENT
.UNINDENT
.sp
In this operation:
.INDENT 0.0
.IP \(bu 2
\fBmongoperf\fP tests direct physical random read and write io\(aqs, using
16 concurrent reader threads.
.IP \(bu 2
\fBmongoperf\fP uses a 10 gigabyte test file.
.UNINDENT
.sp
Consider using \fBiostat\fP, as invoked in the following example to
monitor I/O performance during the test.
.INDENT 0.0
.INDENT 3.5
.sp
.nf
.ft C
iostat \-xtm 1
.ft P
.fi
.UNINDENT
.UNINDENT
.SH AUTHOR
MongoDB Documentation Project
.SH COPYRIGHT
2011-2015
.\" Generated by docutils manpage writer.
.