Monitoring for MongoDB

Monitoring is a critical component of all database administration. A firm grasp of MongoDB’s reporting will allow you to assess the state of your database and maintain your deployment without crisis. Additionally, a sense of MongoDB’s normal operational parameters will allow you to diagnose before they escalate to failures. There are three methods for collecting data about the state of a running MongoDB instance:

First, there is a set of utilities distributed with MongoDB that provides real-time reporting of database activities.
Second, database commands return statistics regarding the current database state with greater fidelity.
Third, MMS Monitoring Service collects data from running MongoDB deployments and provides visualization and alerts based on that data. MMS is a free service provided by MongoDB.

Each strategy can help answer different questions and is useful in different contexts. These methods are complementary.

MongoDB Reporting Tools – This section provides an overview of the reporting methods distributed with MongoDB. It also offers examples of the kinds of questions that each method is best suited to help you address.

Utilities – The MongoDB distribution includes a number of utilities that quickly return statistics about instances’ performance and activity. Typically, these are most useful for diagnosing issues and assessing normal operation.
mongostat – mongostat captures and returns the counts of database operations by type (e.g. insert, query, update, delete, etc.). These counts report on the load distribution on the server. Use mongostat to understand distribution of operation types and to inform capacity planning.
mongotop – mongotop tracks and reports the current read and write activity of a MongoDB instance, and reports these statistics on a per collection basis. Use mongotop to check if your database activity and use match your expectations.
REST Interface – MongoDB provides a simple REST interface that can be useful for configuring monitoring and alert scripts, and for other administrative tasks.
To enable, configure mongod to use REST, either by starting mongod with the –rest option, or by setting the net.http.RESTInterfaceEnabled setting to true in a configuration file.
HTTP Console – MongoDB provides a web interface that exposes diagnostic and monitoring information in a simple web page. The web interface is accessible at localhost:<port>, where the <port> number is 1000 more than the mongod port . For example, if a locally running mongod is using the default port 27017, access the HTTP console at http://localhost:28017.
Commands – MongoDB includes a number of commands that report on the state of the database. These data may provide a finer level of granularity than the utilities discussed above. Consider using their output in scripts and programs to develop custom alerts, or to modify the behavior of your application in response to the activity of your instance. The db.currentOp method is another useful tool for identifying the database instance’s in-progress operations.
serverStatus – The serverStatus command, or db.serverStatus() from the shell, returns a general overview of the status of the database, detailing disk usage, memory use, connection, journaling, and index access. The command returns quickly and does not impact MongoDB performance. serverStatus outputs an account of the state of a MongoDB instance. This command is rarely run directly. In most cases, the data is more meaningful when aggregated, as one would see with monitoring tools including MMS . Nevertheless, all administrators should be familiar with the data provided by serverStatus.
dbStats – The dbStats command, or db.stats() from the shell, returns a document that addresses storage use and data volumes. The dbStats reflect the amount of storage used, the quantity of data contained in the database, and object, collection, and index counters. Use this data to monitor the state and storage capacity of a specific database. This output also allows you to compare use between databases and to determine the average document size in a database.
collStats – The collStats provides statistics that resemble dbStats on the collection level, including a count of the objects in the collection, the size of the collection, the amount of disk space used by the collection, and information about its indexes.
replSetGetStatus – The replSetGetStatus command (rs.status() from the shell) returns an overview of your replica set’s status. The replSetGetStatus document details the state and configuration of the replica set and statistics about its members. Use this data to ensure that replication is properly configured, and to check the connections between the current host and the other members of the replica set.
Third Party Tools – A number of third party monitoring tools have support for MongoDB, either directly, or through their own plugins.

Process Logging – During normal operation, mongod and mongos instances report a live account of all server activity and operations to either standard output or a log file. The following runtime settings control these options.

Limits the amount of information written to the log or output.
Increases the amount of information written to the log or output.
path. Enables logging to a file, rather than the standard output. You must specify the full path to the log file when adjusting this setting.
Adds information to a log file instead of overwriting the file.

You can specify these configuration operations as the command line arguments to mongod or mongos For example:

mongod -v –logpath /var/log/mongodb/server1.log –logappend

Starts a mongod instance in verbose mode, appending data to the log file at /var/log/mongodb/server1.log/. The following database commands also affect logging:

Displays recent messages from the mongod process log.
Rotates the log files for mongod processes only.

Diagnosing Performance Issues – Degraded performance in MongoDB is typically a function of the relationship between the quantity of data stored in the database, the amount of system RAM, the number of connections to the database, and the amount of time the database spends in a locked state.

In some cases performance issues may be transient and related to traffic load, data access patterns, or the availability of hardware on the host system for virtualized environments. Some users also experience performance limitations as a result of inadequate or inappropriate indexing strategies, or as a consequence of poor schema design patterns. In other situations, performance issues may indicate that the database may be operating at capacity and that it is time to add additional capacity to the database. The following are some causes of degraded performance in MongoDB.

Locks – MongoDB uses a locking system to ensure data set consistency. However, if certain operations are long-running, or a queue forms, performance will slow as requests and operations wait for the lock. Lock-related slowdowns can be intermittent. To see if the lock has been affecting your performance, look to the data in the globalLock section of the serverStatus output. If globalLock.currentQueue.total is consistently high, then there is a chance that a large number of requests are waiting for a lock. This indicates a possible concurrency issue that may be affecting performance.

If globalLock.totalTime is high relative to uptime, the database has existed in a lock state for a significant amount of time. If globalLock.ratio is also high, MongoDB has likely been processing a large number of long running queries. Long queries are often the result of a number of factors: ineffective use of indexes, non-optimal schema design, poor query structure, system architecture issues, or insufficient RAM resulting in page faults and disk reads.

Memory Usage – MongoDB uses memory mapped files to store data. Given a data set of sufficient size, the MongoDB process will allocate all available memory on the system for its use. While this is part of the design, and affords MongoDB superior performance, the memory mapped files make it difficult to determine if the amount of RAM is sufficient for the data set.

The memory usage statuses metrics of the serverStatus output can provide insight into MongoDB’s memory use. Check the resident memory use (i.e. mem.resident): if this exceeds the amount of system memory and there is a significant amount of data on disk that isn’t in RAM, you may have exceeded the capacity of your system. You should also check the amount of mapped memory (i.e. mem.mapped.) If this value is greater than the amount of system memory, some operations will require disk access page faults to read data from virtual memory and negatively affect performance.

Page Faults – A page fault occurs when MongoDB requires data not located in physical memory, and must read from virtual memory. To check for page faults, see the extra_info.page_faults value in the serverStatus output. This data is only available on Linux systems. A single page fault completes quickly and is not problematic. However, in aggregate, large volumes of page faults typically indicate that MongoDB is reading too much data from disk. In many situations, MongoDB’s read locks will “yield” after a page fault to allow other processes to read and avoid blocking while waiting for the next page to read into memory. This approach improves concurrency, and also improves overall throughput in high volume systems.

Increasing the amount of RAM accessible to MongoDB may help reduce the number of page faults. If this is not possible, you may want to consider deploying a sharded cluster and/or adding shards to your deployment to distribute load among mongod instances.

Number of Connections – In some cases, the number of connections between the application layer (i.e. clients) and the database can overwhelm the ability of the server to handle requests. This can produce performance irregularities. The following fields in the serverStatus document can provide insight:

activeClients contains a counter of the total number of clients with active operations in progress or queued.
connections is a container for the following two fields:
current the total number of current clients that connect to the database instance.
available the total number of unused collections available for new clients.

If requests are high because there are numerous concurrent application requests, the database may have trouble keeping up with demand. If this is the case, then you will need to increase the capacity of your deployment. For read-heavy applications increase the size of your replica set and distribute read operations to secondary members. For write heavy applications, deploy sharding and add one or more shards to a sharded cluster to distribute load among mongod instances.

Spikes in the number of connections can also be the result of application or driver errors. All of the officially supported MongoDB drivers implement connection pooling, which allows clients to use and reuse connections more efficiently. Extremely high numbers of connections, particularly without corresponding workload is often indicative of a driver or other configuration error.

Unless constrained by system-wide limits MongoDB has no limit on incoming connections. You can modify system limits using the ulimit command, or by editing your system’s /etc/sysctl file.

Database Profiling – MongoDB’s “Profiler” is a database profiling system that can help identify inefficient queries and operations. The following profiling levels are available:

Level	Setting
0	Off. No profiling
1	On. Only includes “slow” operations
2	On. Includes all operations

Enable the profiler by setting the profile value using the following command in the mongo shell:

db.setProfilingLevel(1)

The slowOpThresholdMs setting defines what constitutes a “slow” operation. To set the threshold above which the profiler considers operations “slow” (and thus, included in the level 1 profiling data), you can configure slowOpThresholdMs at runtime as an argument to the db.setProfilingLevel() operation. By default, mongod records all “slow” queries to its log, as defined by slowOpThresholdMs. Unlike log data, the data in system.profile does not persist between mongod restarts. Because the database profiler can negatively impact performance, only enable profiling for strategic intervals and as minimally as possible on production systems. You may enable profiling on a per-mongod basis. This setting will not propagate across a replica set or sharded cluster. You can view the output of the profiler in the system.profile collection of your database by issuing the show profile command in the mongo shell, or with the following operation:

db.system.profile.find( { millis : { $gt : 100 } } )

This returns all operations that lasted longer than 100 milliseconds. Ensure that the value specified here (100, in this example) is above the slowOpThresholdMs threshold.

Replication and Monitoring – Beyond the basic monitoring requirements for any MongoDB instance, for replica sets, administrators must monitor replication lag. “Replication lag” refers to the amount of time that it takes to copy (i.e. replicate) a write operation on the primary to a secondary. Some small delay period may be acceptable, but two significant problems emerge as replication lag grows:

First, operations that occurred during the period of lag are not replicated to one or more secondaries. If you’re using replication to ensure data persistence, exceptionally long delays may impact the integrity of your data set.
Second, if the replication lag exceeds the length of the operation log (oplog) then MongoDB will have to perform an initial sync on the secondary, copying all data from the primary and rebuilding all indexes. This is uncommon under normal circumstances, but if you configure the oplog to be smaller than the default, the issue can arise.

The size of the oplog is only configurable during the first run using the –oplogSize argument to the mongod command, or preferably, the oplogSizeMB setting in the MongoDB configuration file. If you do not specify this on the command line before running with the –replSet option, mongod will create a default sized oplog. By default, the oplog is 5 percent of total available disk space on 64-bit systems. Replication issues are most often the result of network connectivity issues between members, or the result of a primary that does not have the resources to support application and replication traffic. To check the status of a replica, use the replSetGetStatus or the following helper in the shell:

rs.status()

The replSetGetStatus document provides a more in-depth overview view of this output. In general, watch the value of optimeDate, and pay particular attention to the time difference between the primary and the secondary members.