High Availability and PyMongo

PyMongo makes it easy to write highly available applications whether you use a single replica set or a large sharded cluster.

Connecting to a Replica Set

PyMongo makes working with replica sets easy. Here we’ll launch a new replica set and show how to handle both initialization and normal connections with PyMongo.

Note

Replica sets require server version >= 1.6.0. Support for connecting to replica sets also requires PyMongo version >= 1.8.0.

See general MongoDB documentation

rs

Starting a Replica Set

The main replica set documentation contains extensive information about setting up a new replica set or migrating an existing MongoDB setup, be sure to check that out. Here, we’ll just do the bare minimum to get a three node replica set setup locally.

Warning

Replica sets should always use multiple nodes in production - putting all set members on the same physical node is only recommended for testing and development.

We start three mongod processes, each on a different port and with a different dbpath, but all using the same replica set name “foo”. In the example we use the hostname “morton.local”, so replace that with your hostname when running:

$ hostname
morton.local
$ mongod --replSet foo/morton.local:27018,morton.local:27019 --rest
$ mongod --port 27018 --dbpath /data/db1 --replSet foo/morton.local:27017 --rest
$ mongod --port 27019 --dbpath /data/db2 --replSet foo/morton.local:27017 --rest

Initializing the Set

At this point all of our nodes are up and running, but the set has yet to be initialized. Until the set is initialized no node will become the primary, and things are essentially “offline”.

To initialize the set we need to connect to a single node and run the initiate command. Since we don’t have a primary yet, we’ll need to tell PyMongo that it’s okay to connect to a slave/secondary:

>>> from pymongo import MongoClient, ReadPreference
>>> c = MongoClient("morton.local:27017",
                    read_preference=ReadPreference.SECONDARY)

Note

We could have connected to any of the other nodes instead, but only the node we initiate from is allowed to contain any initial data.

After connecting, we run the initiate command to get things started (here we just use an implicit configuration, for more advanced configuration options see the replica set documentation):

>>> c.admin.command("replSetInitiate")
{u'info': u'Config now saved locally.  Should come online in about a minute.',
 u'info2': u'no configuration explicitly specified -- making one', u'ok': 1.0}

The three mongod servers we started earlier will now coordinate and come online as a replica set.

Connecting to a Replica Set

The initial connection as made above is a special case for an uninitialized replica set. Normally we’ll want to connect differently. A connection to a replica set can be made using the normal MongoClient() constructor, specifying one or more members of the set. For example, any of the following will create a connection to the set we just created:

>>> MongoClient("morton.local", replicaset='foo')
MongoClient([u'morton.local:27019', 'morton.local:27017', u'morton.local:27018'])
>>> MongoClient("morton.local:27018", replicaset='foo')
MongoClient([u'morton.local:27019', u'morton.local:27017', 'morton.local:27018'])
>>> MongoClient("morton.local", 27019, replicaset='foo')
MongoClient(['morton.local:27019', u'morton.local:27017', u'morton.local:27018'])
>>> MongoClient(["morton.local:27018", "morton.local:27019"])
MongoClient(['morton.local:27019', u'morton.local:27017', 'morton.local:27018'])
>>> MongoClient("mongodb://morton.local:27017,morton.local:27018,morton.local:27019")
MongoClient(['morton.local:27019', 'morton.local:27017', 'morton.local:27018'])

The nodes passed to MongoClient() are called the seeds. If only one host is specified the replicaset parameter must be used to indicate this isn’t a connection to a single node. As long as at least one of the seeds is online, the driver will be able to “discover” all of the nodes in the set and make a connection to the current primary.

Handling Failover

When a failover occurs, PyMongo will automatically attempt to find the new primary node and perform subsequent operations on that node. This can’t happen completely transparently, however. Here we’ll perform an example failover to illustrate how everything behaves. First, we’ll connect to the replica set and perform a couple of basic operations:

>>> db = MongoClient("morton.local", replicaSet='foo').test
>>> db.test.save({"x": 1})
ObjectId('...')
>>> db.test.find_one()
{u'x': 1, u'_id': ObjectId('...')}

By checking the host and port, we can see that we’re connected to morton.local:27017, which is the current primary:

>>> db.connection.host
'morton.local'
>>> db.connection.port
27017

Now let’s bring down that node and see what happens when we run our query again:

>>> db.test.find_one()
Traceback (most recent call last):
pymongo.errors.AutoReconnect: ...

We get an AutoReconnect exception. This means that the driver was not able to connect to the old primary (which makes sense, as we killed the server), but that it will attempt to automatically reconnect on subsequent operations. When this exception is raised our application code needs to decide whether to retry the operation or to simply continue, accepting the fact that the operation might have failed.

On subsequent attempts to run the query we might continue to see this exception. Eventually, however, the replica set will failover and elect a new primary (this should take a couple of seconds in general). At that point the driver will connect to the new primary and the operation will succeed:

>>> db.test.find_one()
{u'x': 1, u'_id': ObjectId('...')}
>>> db.connection.host
'morton.local'
>>> db.connection.port
27018

MongoReplicaSetClient

Using a MongoReplicaSetClient instead of a simple MongoClient offers two key features: secondary reads and replica set health monitoring. To connect using MongoReplicaSetClient just provide a host:port pair and the name of the replica set:

>>> from pymongo import MongoReplicaSetClient
>>> MongoReplicaSetClient("morton.local:27017", replicaSet='foo')
MongoReplicaSetClient([u'morton.local:27019', u'morton.local:27017', u'morton.local:27018'])

Secondary Reads

By default an instance of MongoReplicaSetClient will only send queries to the primary member of the replica set. To use secondaries for queries we have to change the ReadPreference:

>>> db = MongoReplicaSetClient("morton.local:27017", replicaSet='foo').test
>>> from pymongo.read_preferences import ReadPreference
>>> db.read_preference = ReadPreference.SECONDARY_PREFERRED

Now all queries will be sent to the secondary members of the set. If there are no secondary members the primary will be used as a fallback. If you have queries you would prefer to never send to the primary you can specify that using the SECONDARY read preference:

>>> db.read_preference = ReadPreference.SECONDARY

Read preference can be set on a client, database, collection, or on a per-query basis, e.g.:

>>> db.collection.find_one(read_preference=ReadPreference.PRIMARY)

Reads are configured using three options: read_preference, tag_sets, and secondary_acceptable_latency_ms.

read_preference:

  • PRIMARY: Read from the primary. This is the default, and provides the strongest consistency. If no primary is available, raise AutoReconnect.
  • PRIMARY_PREFERRED: Read from the primary if available, or if there is none, read from a secondary matching your choice of tag_sets and secondary_acceptable_latency_ms.
  • SECONDARY: Read from a secondary matching your choice of tag_sets and secondary_acceptable_latency_ms. If no matching secondary is available, raise AutoReconnect.
  • SECONDARY_PREFERRED: Read from a secondary matching your choice of tag_sets and secondary_acceptable_latency_ms if available, otherwise from primary (regardless of the primary’s tags and latency).
  • NEAREST: Read from any member matching your choice of tag_sets and secondary_acceptable_latency_ms.

tag_sets:

Replica-set members can be tagged according to any criteria you choose. By default, MongoReplicaSetClient ignores tags when choosing a member to read from, but it can be configured with the tag_sets parameter. tag_sets must be a list of dictionaries, each dict providing tag values that the replica set member must match. MongoReplicaSetClient tries each set of tags in turn until it finds a set of tags with at least one matching member. For example, to prefer reads from the New York data center, but fall back to the San Francisco data center, tag your replica set members according to their location and create a MongoReplicaSetClient like so:

>>> rsc = MongoReplicaSetClient(
...     "morton.local:27017",
...     replicaSet='foo'
...     read_preference=ReadPreference.SECONDARY,
...     tag_sets=[{'dc': 'ny'}, {'dc': 'sf'}]
... )

MongoReplicaSetClient tries to find secondaries in New York, then San Francisco, and raises AutoReconnect if none are available. As an additional fallback, specify a final, empty tag set, {}, which means “read from any member that matches the mode, ignoring tags.”

secondary_acceptable_latency_ms:

If multiple members match the mode and tag sets, MongoReplicaSetClient reads from among the nearest members, chosen according to ping time. By default, only members whose ping times are within 15 milliseconds of the nearest are used for queries. You can choose to distribute reads among members with higher latencies by setting secondary_acceptable_latency_ms to a larger number. In that case, MongoReplicaSetClient distributes reads among matching members within secondary_acceptable_latency_ms of the closest member’s ping time.

Note

secondary_acceptable_latency_ms is ignored when talking to a replica set through a mongos. The equivalent is the localThreshold command line option.

Health Monitoring

When MongoReplicaSetClient is initialized it launches a background task to monitor the replica set for changes in:

  • Health: detect when a member goes down or comes up, or if a different member becomes primary
  • Configuration: detect changes in tags
  • Latency: track a moving average of each member’s ping time

Replica-set monitoring ensures queries are continually routed to the proper members as the state of the replica set changes.

It is critical to call close() to terminate the monitoring task before your process exits.

High Availability and mongos

An instance of MongoClient can be configured to automatically connect to a different mongos if the instance it is currently connected to fails. If a failure occurs, PyMongo will attempt to find the nearest mongos to perform subsequent operations. As with a replica set this can’t happen completely transparently, Here we’ll perform an example failover to illustrate how everything behaves. First, we’ll connect to a sharded cluster, using a seed list, and perform a couple of basic operations:

>>> db = MongoClient('morton.local:30000,morton.local:30001,morton.local:30002').test
>>> db.test.save({"x": 1})
ObjectId('...')
>>> db.test.find_one()
{u'x': 1, u'_id': ObjectId('...')}

Each member of the seed list passed to MongoClient must be a mongos. By checking the host, port, and is_mongos attributes we can see that we’re connected to morton.local:30001, a mongos:

>>> db.connection.host
'morton.local'
>>> db.connection.port
30001
>>> db.connection.is_mongos
True

Now let’s shut down that mongos instance and see what happens when we run our query again:

>>> db.test.find_one()
Traceback (most recent call last):
pymongo.errors.AutoReconnect: ...

As in the replica set example earlier in this document, we get an AutoReconnect exception. This means that the driver was not able to connect to the original mongos at port 30001 (which makes sense, since we shut it down), but that it will attempt to connect to a new mongos on subsequent operations. When this exception is raised our application code needs to decide whether to retry the operation or to simply continue, accepting the fact that the operation might have failed.

As long as one of the seed list members is still available the next operation will succeed:

>>> db.test.find_one()
{u'x': 1, u'_id': ObjectId('...')}
>>> db.connection.host
'morton.local'
>>> db.connection.port
30002
>>> db.connection.is_mongos
True