The Zope Object Database provides an object-oriented database for Python that provides a high-degree of transparency. Applications can take advantage of object database features with few, if any, changes to application logic. ZODB includes features such as a plugable storage interface, rich transaction support, and undo.
Contents
The ZODB package provides a set of tools for using the Zope Object Database (ZODB). The components you get with the ZODB release are as follows:
Our primary development platforms are Linux, Mac OS X, and Windows XP. The test suite should pass without error on all of these platforms, although it can take a long time on Windows -- longer if you use ZoneAlarm. Many particularly slow tests are skipped unless you pass --all as an argument to test.py.
ZODB 3.10 requires Python 2.5 or later.
ZODB ZEO clients from ZODB 3.2 on can talk to ZODB 3.10 servers. ZODB ZEO 3.10 Clients can talk to ZODB 3.8, 3.9, and 3.10 ZEO servers.
You must have Python installed. If you're using a system Python install, make sure development support is installed too.
You also need the transaction, zc.lockfile, ZConfig, zdaemon, zope.event, zope.interface, zope.proxy and zope.testing packages. If you don't have them and you can connect to the Python Package Index, then these will be installed for you if you don't have them.
ZODB is released as a distutils package. The easiest ways to build and install it are to use easy_install, or zc.buildout.
To install by hand, first install the dependencies, ZConfig, zdaemon, zope.interface, zope.proxy and zope.testing. These can be found in the Python Package Index.
To run the tests, use the test setup command:
python setup.py test
It will download dependencies if needed. If this happens, ou may get an import error when the test command gets to looking for tests. Try running the test command a second time and you should see the tests run.
python setup.py test
To install, use the install command:
python setup.py install
The ZODB checkouts are buildouts. When working from a ZODB checkout, first run the bootstrap.py script to initialize the buildout:
% python bootstrap.py
and then use the buildout script to build ZODB and gather the dependencies:
% bin/buildout
This creates a test script:
% bin/test -v
This command will run all the tests, printing a single dot for each test. When it finishes, it will print a test summary. The exact number of tests can vary depending on platform and available third-party libraries.:
Ran 1182 tests in 241.269s OK
The test script has many more options. Use the -h or --help options to see a file list of options. The default test suite omits several tests that depend on third-party software or that take a long time to run. To run all the available tests use the --all option. Running all the tests takes much longer.:
Ran 1561 tests in 1461.557s OK
Several scripts are provided with the ZODB and can help for analyzing, debugging, checking for consistency, summarizing content, reporting space used by objects, doing backups, artificial load testing, etc. Look at the ZODB/script directory for more informations.
The historical version numbering schemes for ZODB and ZEO are complicated. Starting with ZODB 3.4, the ZODB and ZEO version numbers are the same.
In the ZODB 3.1 through 3.3 lines, the ZEO version number was "one smaller" than the ZODB version number; e.g., ZODB 3.2.7 included ZEO 2.2.7. ZODB and ZEO were distinct releases prior to ZODB 3.1, and had independent version numbers.
Historically, ZODB was distributed as a part of the Zope application server. Jim Fulton's paper at the Python conference in 2000 described a version of ZODB he called ZODB 3, based on an earlier persistent object system called BoboPOS. The earliest versions of ZODB 3 were released with Zope 2.0.
Andrew Kuchling extracted ZODB from Zope 2.4.1 and packaged it for use by standalone Python programs. He called this version "StandaloneZODB". Andrew's guide to using ZODB is included in the Doc directory. This version of ZODB was hosted at http://sf.net/projects/zodb. It supported Python 1.5.2, and might still be of interest to users of this very old Python version.
Zope Corporation released a version of ZODB called "StandaloneZODB 1.0" in Feb. 2002. This release was based on Andrew's packaging, but built from the same CVS repository as Zope. It is roughly equivalent to the ZODB in Zope 2.5.
Why not call the current release StandaloneZODB? The name StandaloneZODB is a bit of a mouthful. The standalone part of the name suggests that the Zope version is the real version and that this is an afterthought, which isn't the case. So we're calling this release "ZODB". We also worked on a ZODB4 package for a while and made a couple of alpha releases. We've now abandoned that effort, because we didn't have the resources to pursue ot while also maintaining ZODB(3).
ZODB is distributed under the Zope Public License, an OSI-approved open source license. Please see the LICENSE.txt file for terms and conditions.
The ZODB/ZEO Programming Guide included in the documentation is a modified version of Andrew Kuchling's original guide, provided under the terms of the GNU Free Documentation License.
We maintain a Wiki page about all things ZODB, including status on future directions for ZODB. Please see
http://wiki.zope.org/ZODB/FrontPage
and feel free to contribute your comments. There is a Mailman mailing list in place to discuss all issues related to ZODB. You can send questions to
zodb-dev@zope.org
or subscribe at
http://lists.zope.org/mailman/listinfo/zodb-dev
and view its archives at
http://lists.zope.org/pipermail/zodb-dev
Note that Zope Corp mailing lists have a subscriber-only posting policy.
Andrew's ZODB Programmers Guide is made available in several forms, including DVI and HTML. To view it online, point your browser at the file Doc/guide/zodb/index.html
Bug reports and patches should be added to the Launchpad:
https://launchpad.net/zodb
When a transaction rolled back a savepoint after adding objects and subsequently added more objects and committed, an error could be raised "ValueError: A different object already has the same oid" causing the transaction to fail. Worse, this could leave a database in a state where subsequent transactions in the same process would fail.
Unix domain sockets didn't work for ZEO (since the addition of IPv6 support). https://bugs.launchpad.net/zodb/+bug/663259
Removed a missfeature that can cause performance problems when using an external garbage collector with ZEO. When objects were deleted from a storage, invalidations were sent to clients. This makes no sense. It's wildly unlikely that the other connections/clients have copies of the garbage. In normal storage garbage collection, we don't send invalidations. There's no reason to send them when an external garbage collector is used.
ZEO client cache simulation misshandled invalidations causing incorrect statistics and errors.
There are a number of performance enhancements for ZEO storage servers.
FileStorage indexes use a new format. They are saved and loaded much faster and take less space. Old indexes can still be read, but new indexes won't be readable by older versions of ZODB.
The API for undoing multiple transactions has changed. To undo multiple transactions in a single transaction, pass a list of transaction identifiers to a database's undoMultiple method. Calling a database's undo method multiple times in the same transaction now raises an exception.
The ZEO protocol for undo has changed. The only user-visible consequence of this is that when ZODB 3.10 ZEO servers won't support undo for older clients.
The storage API (IStorage) has been tightened. Now, storages should raise a StorageTransactionError when invalid transactions are passed to tpc_begin, tpc_vote, or tpc_finish.
ZEO clients (ClientStorage instances) now work in forked processes, including those created via multiprocessing.Process instances.
Broken objects now provide the IBroken interface.
As a convenience, you can now pass an integer port as an address to the ZEO ClientStorage constructor.
As a convenience, there's a new client function in the ZEO package for constructing a ClientStorage instance. It takes the same arguments as the ClientStorage constructor.
DemoStorages now accept constructor athuments, close_base_on_close and close_changes_on_close, to control whether underlying storages are closed when the DemoStorage is closed.
Removed the dependency on zope.proxy.
Removed support for the _p_independent mini framework, which was made moot by the introduction of multi-version concurrency control several years ago.
Added support for the transaction retry convenience (transaction-manager attempts method) introduced in the transaction 1.1.0 release.
Enhanced the database opening conveniences:
Databases now warn when committing very large records (> 16MB). This is to try to warn people of likely design mistakes. There is a new option (large_record_size/large-record-size) to control the record size at which the warning is issued.
Added support for wrapper storages that transform pickle data. Applications for this include compression and encryption. An example wrapper storage implementation, ZODB.tests.hexstorage, was included for testing.
It is important that storage implementations not assume that storages contain pickles. Renamed IStorageDB to IStorageWrapper and expanded it to provide methods for transforming and untransforming data records. Storages implementations should use these methods to get pickle data from stored records.
Deprecated ZODB.interfaces.StorageStopIteration. Storage iterator implementations should just raise StopIteration, which means they can now be implemented as generators.
The filestorage packer configuration option noe accepts values of the form modname:expression, allowing the use of packer factories with options.
Added a new API that allows applications to make sure that current data are read. For example, with:
self._p_jar.readCurrent(ob)
A conflict error will be raised if the version of ob read by the transaction isn't current when the transaction is committed.
Normally, ZODB only assures that objects read are consistent, but not necessarily up to date. Checking whether an object is up to date is important when information read from one object is used to update another.
BTrees are an important case of reading one object to update another. Internal nodes are read to decide which leave notes are updated when a BTree is updated. BTrees now use this new API to make sure that internal nodes are up to date on updates.
When transactions are aborted, new object ids allocated during the transaction are saved and used in subsequent transactions. This can help in situations where object ids are used as BTree keys and the sequential allocation of object ids leads to conflict errors.
ZEO servers now support a server_status method for for getting information on the number of clients, lock requests and general statistics.
ZEO clients now support a client_label constructor argument and client-label configuration-file option to specify a label for a client in server logs. This makes it easier to identify specific clients corresponding to server log entries, especially when there are multiple clients originating from the same machine.
Improved ZEO server commit lock logging. Now, locking activity is logged at the debug level until the number of waiting lock requests gets above 3. Log at the critical level when the number of waiting lock requests gets above 9.
The file-storage backup script, repozo, will now create a backup index file if an output file name is given via the --output/-o option.
Added a '--kill-old-on-full' argument to the repozo backup options: if passed, remove any older full or incremental backup files from the repository after doing a full backup. (https://bugs.launchpad.net/zope2/+bug/143158)
The mkzeoinst script has been moved to a separate project:
and is no-longer included with ZODB.
Removed untested unsupported dbmstorage fossile.
ZEO servers no longer log their pids in every log message. It's just not interesting. :)
When a pool timeout was specified for a database and old connections were removed due to timing out, an error occured due to a bug in the connection cleanup logic.
When multi-database connections were no longer used and cleaned up, their subconnections weren't cleaned up properly.
ZEO didn't work with IPv6 addrsses. Added IPv6 support contributed by Martin v. Löwis.
A file storage bug could cause ZEO clients to have incorrect information about current object revisions after reconnecting to a database server.
Updated the 'repozo --kill-old-on-full' option to remove any '.index' files corresponding to backups being removed.
ZEO extension methods failed when a client reconnected to a storage. (https://bugs.launchpad.net/zodb/+bug/143344)
Clarified the return Value for lastTransaction in the case when there aren't any transactions. Now a string of 8 nulls (aka "z64") is specified.
Setting _p_changed on a blob wo actually writing anything caused an error. (https://bugs.launchpad.net/zodb/+bug/440234)
The verbose mode of the fstest was broken. (https://bugs.launchpad.net/zodb/+bug/475996)
Object ids created in a savepoint that is rolled back wren't being reused. (https://bugs.launchpad.net/zodb/+bug/588389)
Database connections didn't invalidate cache entries when conflict errors were raised in response to checkCurrentSerialInTransaction errors. Normally, this shouldn't be a problem, since there should be pending invalidations for these oids which will cause the object to be invalidated. There have been issues with ZEO persistent cache management that have caused out of date data to remain in the cache. (It's possible that the last of these were addressed in the 3.10.0b5.) Invalidating read data when there is a conflict error provides some extra insurance.
The interface, ZODB.interfaces.IStorage was incorrect. The store method should never return a sequence of oid and serial pairs.
When a demo storage push method was used to create a new demo storage and the new storage was closed, the original was (incorrectly) closed.
There were numerous bugs in the ZEO cache tracing and analysis code. Cache simulation, while not perfect, seems to be much more accurate now than it was before.
The ZEO cache trace statistics and simulation scripts have been given more descriptive names and moved to the ZEO scripts package.
BTree sets and tree sets didn't correctly check values passed to update or to constructors, causing Python to exit under certain circumstances.
Fixed bug in copying a BTrees.Length instance. (https://bugs.launchpad.net/zodb/+bug/516653)
Fixed a serious bug that caused cache failures when run with Python optimization turned on.
When using using a ClientStorage in a Storage server, there was a threading bug that caused clients to get disconnected.
On Mac OS X, clients that connected and disconnected quickly could cause a ZEO server to stop accepting connections, due to a failure to catch errors in the initial part of the connection process.
The failure to properly handle exceptions while accepting connections is potentially problematic on other platforms.
Object state management wasn't done correctly when classes implemented custom _p_deavtivate methods. (https://bugs.launchpad.net/zodb/+bug/185066)