jaraco.mongodb 3.13

Routines and classes supporting MongoDB environments

jaraco.mongodb

Documentation

Provides support for MongoDB environments.

sessions

jaraco.mongodb.sessions implements a CherryPy Sessions store backed by MongoDB.

By default, the session store will handle sessions with any objects that can be inserted into a MongoDB collection naturally.

To support richer objects, one may configure the codec to use jaraco.modb.

monitor-index-creation

To monitor an ongoing index operation in a server, simply invoke:

python -m jaraco.mongodb.monitor-index-creation mongodb://host/db

oplog

This package provides an oplog module, which is based on the mongooplog-alt project, which itself is a Python remake of official mongooplog utility, shipped with MongoDB starting from version 2.2.0. It reads oplog of a remote server, and applies operations to the local server. This can be used to keep independed replica set loosly synced in a sort of one way replication, and may be useful in various backup and migration scenarios.

oplog implements basic functionality of the official utility and adds following features:

• tailable oplog reader: runs forever polling new oplog event which is extremly useful for keeping two independent replica sets in almost real-time sync.

• option to sync only selected databases/collections.

• option to exclude one or more namespaces (i.e. dbs or collections) from being synced.

• ability to “rename” dbs/collections on fly, i.e. destination namespaces can differ from the original ones.

• works on mongodb 1.8.x, 2.0.x, and 2.2.x. Official utility only supports version 2.2.x and higher.

• save last processed timestamp to file, resume from saved point later.

• at the time of writing (2.2.0), official mongooplog suffers from bug that limits its usage with replica sets (https://jira.mongodb.org/browse/SERVER-6915)

Invoke the command as a module script: python -m jaraco.mongodb.oplog.

Command-line options

Options common to original mongooplog:

--from <hostname><:port>
  Hostname of the mongod server from which oplog operations are going to be
  pulled.

--host <hostname><:port>, -h

  Hostname of the mongod server to which oplog operations are going to be
  applied. Default is "localhost"

--port <number>

  Port of the mongod server to which oplog operations are going to be
  applied, if not specified in ``--host``. Default is 27017.

-s SECONDS, --seconds SECONDS

  seconds to go back. If not set, try read timestamp from --resume-file.
  If the file not found, assume --seconds=86400 (24 hours)

Options specific to this implementation:

--to
  An alias for ``--host``.

--follow, -f

  Wait for new data in oplog. Makes the utility polling oplog forever (until
  interrupted). New data is going to be applied immideately with at most one
  second delay.

--exclude, -x

   List of space separated namespaces which should be ignored. Can be in form
   of ``dname`` or ``dbname.collection``.

 --ns

   Process only these namespaces, ignoring all others. Space separated list of
   strings in form of ``dname`` or ``dbname.collection``.

 --rename [ns_old=ns_new [ns_old=ns_new ...]]

   Rename database(s) and/or collection(s). Operations on namespace ``ns_old``
   from the source server will be applied to namespace ``ns_new`` on the
   destination server.

 --resume-file FILENAME

   resume from timestamp read from this file and write last processed
   timestamp back to this file (default is mongooplog.ts).
   Pass empty string or 'none' to disable this feature.

Example usages

Consider the following sample usage:

python -m jaraco.mongodb.oplog --from prod.example.com:28000 --to dev.example.com:28500 -f --exclude logdb data.transactions --seconds 600

This command is going to take operations from the last 10 minutes from prod, and apply them to dev. Database logdb and collection transactions of data database will be omitted. After operations for the last minutes will be applied, command will wait for new changes to come, keep running until Ctrl+C or other termination signal recieved.

Testing

Tests for oplog are written in javascript using test harness which is used for testing MongoDB iteself. You can run the whole suite with:

mongo tests/suite.js

Note, that you will need existing writable /data/db dir.

Tests produce alot of output. Succesful execution ends with line like this:

ReplSetTest stopSet *** Shut down repl set - test worked ****