Permalink
Commits on Aug 17, 2010
  1. * Set RELEASE to 2.0-alpha1

    steveschnepp committed Aug 17, 2010
  2. Releasing the 2.0 alpha 1

    steveschnepp committed Aug 17, 2010
Commits on Aug 10, 2010
  1. typo

    steveschnepp committed Aug 10, 2010
Commits on Aug 7, 2010
  1. The timestamp isn't part of the multigraph protocol.

    It's in spooldir since it enables a fast seeking
    steveschnepp committed Aug 7, 2010
Commits on Aug 1, 2010
Commits on Jul 29, 2010
  1. can't 'last' or 'next' out of a subroutine. so check for wait() error…

    …s back where it was. and if there was an error, no further action should be attempted, so restart the loop.
    Matthew Boyle committed Jul 29, 2010
Commits on Jul 28, 2010
  1. move the code for restarting crashed pollers into it's own method. it…

    … should be just about testable that way :-)
    
    update the pid -> service hash as part of the poller launch process.
    Matthew Boyle committed Jul 28, 2010
  2. remove test code that's no longer needed.

    Matthew Boyle committed Jul 28, 2010
  3. get the intervals in _launch_pollers(), and the node list in _get_ser…

    …vice_list().
    
    return a reference to the intervals for convenience.
    Matthew Boyle committed Jul 28, 2010
  4. do the same for the intervals.

    Matthew Boyle committed Jul 28, 2010
  5. stuff the pid -> service mapping into the instance.

    Matthew Boyle committed Jul 28, 2010
  6. remove the pid file in the signal handler, rather than in the destruc…

    …tor. the pollers also call the destructor...
    Matthew Boyle committed Jul 28, 2010
  7. check the node in run() rather than new(). that way the tests don't a…

    …ll break.
    Matthew Boyle committed Jul 28, 2010
  8. keep a track of when each poller was last restarted, and give up on t…

    …hem if it was too recent (currently <10s). should prevent a critically broken service from crashing over and over again.
    
    maybe 10s should really be $service_interval...
    Matthew Boyle committed Jul 28, 2010
  9. remove the pid file on exit.

    Matthew Boyle committed Jul 28, 2010
Commits on Jul 27, 2010
  1. add an "epoch" to the filename. it's not currently used, just intende…

    …d for future-compatibility.
    Matthew Boyle committed Jul 27, 2010
  2. tidy the comments.

    Matthew Boyle committed Jul 27, 2010
  3. _cat_file() is no longer needed.

    Matthew Boyle committed Jul 27, 2010
  4. slurp the rest of the file in one go, rather than concatenating it li…

    …ne-by-line.
    Matthew Boyle committed Jul 27, 2010
  5. tidy up so all the tests pass once more.

    Matthew Boyle committed Jul 27, 2010
  6. fetch doesn't return results for services in any particular order, so…

    … make sure the tests allow for this.
    Matthew Boyle committed Jul 27, 2010
  7. s/$plugin/$service/g

    Matthew Boyle committed Jul 27, 2010
  8. read-side of the change in r3828 (write out the timestamp at the star…

    …t of each set of results, rather than munging the data).
    Matthew Boyle committed Jul 27, 2010
  9. filenames have changed.

    Matthew Boyle committed Jul 27, 2010
  10. non-multigraphs should get a multigraph header, so the master knows w…

    …hat it's looking at.
    Matthew Boyle committed Jul 27, 2010
  11. write out the timestamp at the start of each set of results, rather t…

    …han munging the data.
    Matthew Boyle committed Jul 27, 2010
Commits on Jul 26, 2010
  1. update the spoolwriter, based on discussion with Steve Schnepp. all t…

    …he results now go in one file.
    Matthew Boyle committed Jul 26, 2010
  2. rationalise the test code.

    Matthew Boyle committed Jul 26, 2010
  3. check the return from the open and print calls, and log errors if som…

    …ething went wrong.
    Matthew Boyle committed Jul 26, 2010
  4. fix a couple of broken tests the old-fashioned way.

    Matthew Boyle committed Jul 26, 2010
Commits on Jul 25, 2010