UNIX ulimit Settings for Self-Managed Deployments

mongod

• 1 file descriptor for each data file in use by the mongod instance.

• 1 file descriptor for each journal file used by the mongod instance.

• In replica sets, each mongod maintains a connection to all other members of the set.

mongod uses background threads for a number of internal processes, including TTL collections, replication, and replica set health checks, which may require a small number of additional resources.

mongos

In addition to the threads and file descriptors for client connections, mongos must maintain connections to all config servers and all shards, which includes all members of all replica sets.

For mongos, consider the following behaviors:

• mongos instances maintain a connection pool to each shard so that the mongos can reuse connections and quickly fulfill requests without needing to create new connections.

• You can limit the number of incoming connections using the net.maxIncomingConnections run-time option. By restricting the number of incoming connections you can prevent a cascade effect where the mongos creates too many connections on the mongod instances.

ulimit

You can use the ulimit command at the system prompt to check system limits, as in the following example:

$ ulimit -a
-t: cpu time (seconds)         unlimited
-f: file size (blocks)         unlimited
-d: data seg size (kbytes)     unlimited
-s: stack size (kbytes)        8192
-c: core file size (blocks)    0
-m: resident set size (kbytes) unlimited
-u: processes                  64000
-n: file descriptors           64000
-l: locked-in-memory size (kb) unlimited
-v: address space (kb)         unlimited
-x: file locks                 unlimited
-i: pending signals            192276
-q: bytes in POSIX msg queues  819200
-e: max nice                   30
-r: max rt priority            65
-N 15:                         unlimited

ulimit refers to the per-user limitations for various resources. Therefore, if your mongod instance executes as a user that is also running multiple processes, or multiple mongod processes, you might see contention for these resources. Also, be aware that the processes value (i.e. -u) refers to the combined number of distinct processes and sub-process threads.

On Linux, you can change ulimit settings by issuing a command in the following form:

ulimit -n <value>

There are both "hard" and the "soft" ulimit s that affect MongoDB's performance. The "hard" ulimit refers to the maximum number of processes that a user can have active at any time. This is the ceiling: no non-root process can increase the "hard" ulimit. In contrast, the "soft" ulimit is the limit that is actually enforced for a session or process, but any process can increase it up to "hard" ulimit maximum.

A low "soft" ulimit can cause can't create new thread, closing connection errors if the number of connections grows too high. For this reason, it is extremely important to set both ulimit values to the recommended values.

ulimit will modify both "hard" and "soft" values unless the -H or -S modifiers are specified when modifying limit values.

For many distributions of Linux you can change values by substituting the -n option for any possible value in the output of ulimit -a.

After changing the ulimit settings, you must restart the process to take advantage of the modified settings. On Linux, you can use the /proc file system to see the current limitations on a running process.

Depending on your system's configuration, and default settings, any change to system limits made using ulimit may revert following a system restart. Check your distribution and operating system documentation for more information.

You should typically start mongod using systemctl, which uses the ulimit settings:

systemctl start mongod.service

If you do not start mongod using systemctl, systemd overrides some of the ulimit settings. For example, if you start mongod as shown in the following command, then a user slice (such as user-1000.slice) systemd setting is used:

mongod --config ~/mongod.conf

Recommended ulimit Settings

Every deployment may have unique requirements and settings; however, the following thresholds and settings are particularly important for mongod and mongos deployments:

• -f (file size): unlimited

• -t (cpu time): unlimited

• -v (virtual memory): unlimited [1]

• -l (locked-in-memory size): unlimited

• -n (open files): 64000

• -m (memory size): unlimited [1] [2]

• -u (processes/threads): 64000

Always remember to restart your mongod and mongos instances after changing the ulimit settings to ensure that the changes take effect.

Linux distributions using Upstart

For Linux distributions that use Upstart, you can specify limits within service scripts if you start mongod and/or mongos instances as Upstart services. You can do this by using limit stanzas.

Specify the Recommended ulimit Settings, as in the following example:

limit fsize unlimited unlimited    # (file size)
limit cpu unlimited unlimited      # (cpu time)
limit as unlimited unlimited       # (virtual memory size)
limit memlock unlimited unlimited  # (locked-in-memory size)
limit nofile 64000 64000           # (open files)
limit nproc 64000 64000            # (processes/threads)

Each limit stanza sets the "soft" limit to the first value specified and the "hard" limit to the second.

After changing limit stanzas, ensure that the changes take effect by restarting the application services, using the following form:

restart <service name>

Linux distributions using systemd

If you start a mongod and/or mongos instance as a systemd service, you can specify limits within the [Service] section of its service file. The service file has a location like /etc/systemd/system/<process-name>.service.

You can set limits by using resource limit directives.

Specify the Recommended ulimit Settings, as in the following example:

[Service]
# Other directives omitted
# (file size)
LimitFSIZE=infinity
# (cpu time)
LimitCPU=infinity
# (virtual memory size)
LimitAS=infinity
# (locked-in-memory size)
LimitMEMLOCK=infinity
# (open files)
LimitNOFILE=64000
# (processes/threads)
LimitNPROC=64000

Each systemd limit directive sets both the "hard" and "soft" limits to the value specified.

After changing limit stanzas, ensure that the changes take effect by restarting the application services, using the following form:

systemctl restart <service name>

/proc File System

The /proc file-system stores the per-process limits in the file system object located at /proc/<pid>/limits, where <pid> is the process's PID or process identifier. You can use the following bash function to return the content of the limits object for a process or processes with a given name:

return-limits(){
     for process in $@; do
          process_pids=`ps -C $process -o pid --no-headers | cut -d " " -f 2`
          if [ -z $@ ]; then
             echo "[no $process running]"
          else
             for pid in $process_pids; do
                   echo "[$process #$pid -- limits]"
                   cat /proc/$pid/limits
             done
          fi
     done
}

You can copy and paste this function into a current shell session or load it as part of a script. Call the function with one the following invocations:

return-limits mongod
return-limits mongos
return-limits mongod mongos