NFS-safe file locking with timeouts for POSIX systems.
The `flufl.lock` library provides an NFS-safe file-based locking algorithm
influenced by the GNU/Linux `open(2)` manpage, under the description of the
[...] O_EXCL is broken on NFS file systems, programs which rely on it
for performing locking tasks will contain a race condition. The
solution for performing atomic file locking using a lockfile is to
create a unique file on the same fs (e.g., incorporating hostname and
pid), use link(2) to make a link to the lockfile. If link() returns
0, the lock is successful. Otherwise, use stat(2) on the unique file
to check if its link count has increased to 2, in which case the lock
is also successful.
The assumption made here is that there will be no *outside interference*,
e.g. no agent external to this code will ever `link()` to the specific lock
Lock objects support lock-breaking so that you can't wedge a process forever.
This is especially helpful in a web environment, but may not be appropriate
for all applications.
Locks have a *lifetime*, which is the maximum length of time the process
expects to retain the lock. It is important to pick a good number here
because other processes will not break an existing lock until the expected
lifetime has expired. Too long and other processes will hang; too short and
you'll end up trampling on existing process locks -- and possibly corrupting
data. In a distributed (NFS) environment, you also need to make sure that
your clocks are properly synchronized.
`flufl.lock` is Copyright (C) 2007-2015 Barry Warsaw <firstname.lastname@example.org>
Licensed under the terms of the GNU Lesser General Public License, version 3
or later. See the COPYING.LESSER file for details.
* Project home: https://gitlab.com/warsaw/flufl.lock
* Report bugs at: https://gitlab.com/warsaw/flufl.lock/issues
* Code hosting: email@example.com:warsaw/flufl.lock.git
* Documentation: http://flufllock.readthedocs.org/
NEWS for flufl.lock
* Fix the MANIFEST.in so that tox.ini is included in the sdist.
* Drop Python 2.6 compatibility.
* Add Python 3.5 compatibility.
* Include MANIFEST.in in the sdist tarball, otherwise the Debian package
won't built correctly.
* Fix documentation bug. (LP: #1026403)
* Catch ESTALE along with ENOENT, as NFS servers are supposed to (but don't
always) throw ESTALE instead of ENOENT. (LP: #977999)
* Purge all references to `distribute`. (LP: #1263794)
* Add classifiers to setup.py and make the long description more compatible
with the Cheeseshop.
* Other changes to make the Cheeseshop page look nicer. (LP: #680136)
* setup_helper.py version 2.1.
* Support Python 3 without the use of 2to3.
* Make the documentation clear that the `flufl.test.subproc` functions are
not part of the public API. (LP: #838338)
* Fix claim file format in `take_possession()`. (LP: #872096)
* Provide a new API for dealing with possible additional unexpected errnos
while trying to read the lock file. These can happen in some NFS
environments. If you want to retry the read, set the lock file's
`retry_errnos` property to a sequence of errnos. If one of those errnos
occurs, the read is unconditionally (and infinitely) retried.
`retry_errnos` is a property which must be set to a sequence; it has a
getter and a deleter too. (LP: #882261)
* Fixed TypeError in .lock() method due to race condition in _releasetime
property. Found by Stephen A. Goss. (LP: #827052)
* Added lock.details.
* Small adjustment to doctest.
* Add missing exception to __all__.
* Package renamed to flufl.lock.
Try `bzr log lp:flufl.lock` for details.
TODO: Brief introduction on what you do with files - including link to relevant help section.