You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Increased CURRENT_NEB_API_VERSION, you have to recompile your NEB modules. check_result gained an object_ptr member; a module that submits check results may set it to the host/service the result belongs to, or leave it NULL to have Naemon look the object up by name as before.
The command file worker no longer runs as a separate process. The external command file is now polled directly by the main process event loop, so there is only one Naemon daemon process instead of an extra command worker.
Features
New NEBTYPE_HOSTSTATUS_SCHEDULE / NEBTYPE_SERVICESTATUS_SCHEDULE, sent when only the next check time changed, so modules can tell them apart from real status updates (#162)
The timeout field of check result spool files is now read, falling back to the object's configured timeout when the spool entry does not carry a valid value (#539)
Performance improvements for the main event loop.
status.dat, retention.dat and objects.cache are written faster.
Changed
Naemon refuses to start or reload when the retention file exists but cannot be read completely.
allow_empty_hostgroup_assignment had a different default in code and sample config and could not be set to its code default of 2. The code and the sample config now agree on 2 (allow empty hostgroups, show a warning), 2 can be set in the config, and invalid values are rejected (#556)
A warning is now logged when trailing characters in a boolean object directive are ignored; the leading 0 or 1 remains authoritative for compatibility
Bugfixes
nsock_unix() rejects socket paths that do not fit in sockaddr_un.sun_path instead of overrunning the caller's stack; a query_socket path of 108 characters or more crashed Naemon during startup
Reading a file that shrank or failed while being read crashed Naemon with a bus error, for example a check result spool file rewritten by another program
Expired check result files were not removed from check_result_path
The perfdata file was opened even when no perfdata_file was configured, and the "could not be opened" warning was logged when the file opened successfully
A retained next_check scheduled at startup could land a second late because the delay and the new check time used different clocks; acknowledgement expirations that fired up to a millisecond early were dropped instead of retried and never removed the acknowledgement
check_result->timeout was not initialized, so check results could propagate an uninitialized value to the broker (#539)