Skip to content
Commits on Apr 10, 2012
  1. Deserved comments

    Jonathan Steinert committed Apr 10, 2012
  2. Refactor server parts into a class too.

    Jonathan Steinert committed Apr 10, 2012
  3. This debug option didn't really do much

    Jonathan Steinert committed Apr 10, 2012
  4. I'm sorry to whoever needed the LOAD command, it's gone now

    Jonathan Steinert committed Apr 10, 2012
  5. Fix warnings and fatals caused by strict and warnings

    Jonathan Steinert committed Apr 10, 2012
  6. Turn on strict and warnings everywhere

    Jonathan Steinert committed Apr 10, 2012
  7. Split the server classes into package files.

    Jonathan Steinert committed Apr 10, 2012
Commits on Oct 24, 2007
  1. Add specfile

    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@43 75bc8fc0-0210-0410-8e5e-a32055405bc5
    hachi committed Oct 24, 2007
Commits on Jun 6, 2007
  1. - Added a new LOAD command that returns the number

      of trylocks in the last 5 minutes.
    - Added a LOAD line in the STATUS command.
    
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@42 75bc8fc0-0210-0410-8e5e-a32055405bc5
    ykerherve committed Jun 6, 2007
Commits on Jun 5, 2007
  1. Add a 'STATUS' command to the locking server, gives information about…

    … times, how many requests have been served successfully and have failed. Use for monitoring.
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@41 75bc8fc0-0210-0410-8e5e-a32055405bc5
    hachi committed Jun 5, 2007
Commits on Aug 29, 2006
  1. fix the 'releaselock' command to not crash the server. introduced

    when jonathan did pluggable locking backends.  fortunately we never
    actually use the release command.  but we will....
    
    fix the client to use persistent connections to lock servers, caching
    them in the parent client objects.  so lock objects now have up
    references to their client, and it asks for sockets.  when give one,
    it caches that it has a lock on "10.0.0.34:34" only, not the actual
    socket, and gets it back later on DESTROY to unlock it.
    
    this will waste less ephemeral ports and syndication will catch back up.
    
    
    
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@33 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed Aug 29, 2006
Commits on May 30, 2006
  1. author/copyright/license cleanup. and don't always use DBI, just for …

    …DBI backend
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@32 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed May 30, 2006
  2. whitespace only

    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@31 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed May 30, 2006
Commits on Apr 20, 2006
  1. r180@miyako: hachi | 2006-04-20 15:16:47 -0700

     Initial support for DBI based locking (mysql cluster in particular)
    
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@30 75bc8fc0-0210-0410-8e5e-a32055405bc5
    hachi committed Apr 20, 2006
  2. r105@miyako: hachi | 2006-04-13 13:48:41 -0700

     Object calls work better.
    
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@29 75bc8fc0-0210-0410-8e5e-a32055405bc5
    hachi committed Apr 20, 2006
  3. r29@miyako: hachi | 2006-03-29 19:17:35 -0800

     Untested changes to make ddlockd work with dlmfs. A known-working/tested version will be pushed a few seconds later.
    
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@28 75bc8fc0-0210-0410-8e5e-a32055405bc5
    hachi committed Apr 20, 2006
Commits on Aug 8, 2004
  1. add read_buf member. (not in danga::socket)

    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@26 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed Aug 8, 2004
Commits on May 30, 2004
  1. make accepted connections non-blocking

    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@24 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed May 30, 2004
  2. shouldn't be a blocking accept. (not that it matters much since we

    poll on it, but i noticed it when working with other buggy code)
    
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@23 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed May 30, 2004
  3. ignore sigpipe

    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@22 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed May 30, 2004
Commits on May 28, 2004
  1. needs posix for daemonization

    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@18 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed May 28, 2004
Commits on May 27, 2004
  1. - Fixed the -p/--port command-line option to take an integer argument…

    … instead of
    
      being a boolean.
    
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@9 75bc8fc0-0210-0410-8e5e-a32055405bc5
    deveiant committed May 27, 2004
Commits on May 26, 2004
  1. - Added a command-line option for disabling TCP-level keepalives. Not…

    … suitable
    
      for production, but good for developing on non-Linux machines for now.
    
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@6 75bc8fc0-0210-0410-8e5e-a32055405bc5
    deveiant committed May 26, 2004
Commits on May 19, 2004
  1. wrote sockopt level on the keepaliave

    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@4 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed May 19, 2004
  2. heh, not in that package

    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@3 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed May 19, 2004
  3. danga distributed lock daemon

    take 1
    
    based mostly on mogilefs code, which was based largely on perlbal
    code, both of which use Danga::Socket.  so writing this was cake.
    
    currently it has commands for:
    
    * trylock(lockname)  tries to get lock, returns error immediately
          if couldn't get it.  clients can take as many locks as they want
    
    * releaselock(lockname)   but locks are automatically unlocked
           when client disconnects
    
    * locks()  shows existing locks.  mostly for debugging.
    
    * noop()    just to keep connection alive, if clients want it.  not
                actually necessary, since on Linux uses tcp keep-alives
    
    
    NOTE:  all this is useless without a good multi-lock-daemon-aware client.
    that's coming next.  (don't want the lock daemon being a single POF)
    
    this will eventually be used for all LJ's distributed locking.
    
    
    git-svn-id: http://code.sixapart.com/svn/ddlockd/trunk@2 75bc8fc0-0210-0410-8e5e-a32055405bc5
    bradfitz committed May 19, 2004
Something went wrong with that request. Please try again.