[naemon-users] Announcement: 1.0.4 released

[Sven Nierlein]

Hi,

Today we released the version 1.0.4 of naemon-core and naemon-livestatus. Thanks to all
people who were involved in this release.

Here is a summary of what has changed:

Features
  - Downtimes can now be scheduled to start before their time of their scheduling
  - Naemon can now start without any configured objects
  - Add new NEB callback API version with better support for communicating
    cancellation and override reasons
  - Naemon now consistently logs when a notification is suppressed, along with
    the reason why (very useful for troubleshooting notification issues!). This
    feature is enabled by default, and can be overridden by setting
    enable_notification_suppression_reason_logging=0 in naemon.cfg.
  - Scheduling logic completely overhauled to not depend on wallclock time,
    which makes it much more reliable (no more lost or orphaned events!)
  - It is now possible to disable status data updates completely
  (status_update_interval=0)
  - GLib messages are now logged to the regular outputs.

Bugfixes
  - Fix segfault which would happen when a plugin's output is one or more
    newlines only
  - Shadownaemon: fix segfault if plugin output contains csv separator
  - Fix an ancient bug where reconstructing a downtime from retention data
    would result in double downtime alerts being emitted
  - Fix another ancient race condition where a rapidly reload and starting
    naemon would leave you with two instances running
  - Stop logging if check_result_path (deprecated) is not available even
    if it's set
  - Strings in retention data are now properly escaped, fixing various bugs
  - Various init script fixes
  - Fix a race condition that would sometimes lead to complete deadlock
  - Fix a bug where Naemon would allow duplicate objects to be created
  - Replace in-tree (ha-ha) rbtree with GLib's GTree, to get rid of incorrect
    node removal behaviour in the former
  - Fix a race condition that would sometimes occur during a reload
  - Fix a bug in signal handling code which would sometimes cause the command
    file worker to become defunct
  - Fix a bug where the specified retry_interval of a service or host
    would not be properly respected, and the check_interval would be used instead
  - Fix a potential crash (segfault) when freeing parent/child references
  - Fix a bug where "stale" objects would only have one freshness check executed
  - Fix improper handling of CHECK_OPTION_FORCE_EXECUTION, which would lead to
    checks not being run, despite being forcibly scheduled
  - Fix a bug where Naemon wouldn't properly initialize its lock file, but
    start regardless

Other
  - Grafted Nagios git history onto Naemon's for easier code archeology
  - Lots of code cleanup, typo fixes, new test cases, new assertions &
    error handling, fixed memory leaks, logging and refactoring
  - Removed an undocumented, experimental, untested "load control" feature
  - Decoupled Thruk packages, we now use the official upstream Thruk packages

New contributors since v1.0.3:
  Carl Helmertz (chelmertz at op5.com)
  Emil Hessman (emil at hessman.se)
  Robin Hagman (rhagman at op5.com)
  Philip Eklöf (peklof at op5.com)
  Simen Aasland (simen at opera.com)
  Tobias Sjöndin (tsjondin at op5.com)

Packages and the repository have been updated already. Please report problems and
issues on this list or on the github issue tracker.

Cheers,
 Sven