Mongostat ; A Nifty Tool for Mongo DBA

One of the main Mongodb DBA's task is to monitor the usage of Mongodb system and it's load distribution. This could be needed for proactive monitoring, troubleshooting during performance degradation, root cause analysis, or capacity planning.

Mongostat is a nifty tool which comes out of the box with Mongodb which provides wealth of information in a nicely and familiar formatted way. If you have used vmstat, iostat etc on Linux; Mongostat should seem very familiar.


Mongostat dishes out statistics like counts of database operations by type (e.g. insert, query, update, delete, getmore). The vsize column  in Mongostat output shows the amount of virtual memory in megabytes used by the process. There are other very useful columns regarding network traffic, connections, queuing etc.


Following are some of the examples of running mongostat.

[mongo@mongotest data]$ mongostat

insert query update delete getmore command flushes mapped  vsize    res faults qr|qw ar|aw netIn netOut conn     time
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:29
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:30
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:31
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:32
*0    *0     *0     *0       0     2|0       0 160.0M 646.0M 131.0M      0   0|0   0|0  133b    10k    1 12:47:33

*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:34
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:35
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:36
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:37
*0    *0     *0     *0       0     2|0       0 160.0M 646.0M 131.0M      0   0|0   0|0  133b    10k    1 12:47:38

Following displayes just 5 rows of output.

[mongo@mongotest data]$ mongostat -n 5
insert query update delete getmore command flushes mapped  vsize    res faults qr|qw ar|aw netIn netOut conn     time

*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:45
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:46
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:47
*0    *0     *0     *0       0     1|0       0 160.0M 646.0M 131.0M      0   0|0   0|0   79b    10k    1 12:47:48
*0    *0     *0     *0       0     2|0       0 160.0M 646.0M 131.0M      0   0|0   0|0  133b    10k    1 12:47:49

In order to see full list of options:

[mongo@mongotest data]$ mongostat --help

Usage:
mongostat

Monitor basic MongoDB server statistics.

See http://docs.mongodb.org/manual/reference/program/mongostat/ for more information.

general options:

--help                     print usage
--version                  print the tool version and exit
verbosity options:
-v, --verbose                  more detailed log output (include multiple times for more verbosity, e.g. -vvvvv)
--quiet                    hide all log output
connection options:
-h, --host=                    mongodb host to connect to (setname/host1,host2 for replica sets)
--port=                    server port (can also use --host hostname:port)
authentication options:
-u, --username=                username for authentication
-p, --password=                password for authentication
--authenticationDatabase=  database that holds the user's credentials
--authenticationMechanism= authentication mechanism to use
stat options:
--noheaders                don't output column names
-n, --rowcount=                number of stats lines to print (0 for indefinite)
--discover                 discover nodes and display stats for all
--http                     use HTTP instead of raw db connection
--all                      all optional fields
--json                     output as JSON rather than a formatted table