Centreon Engine 19.10.13¶
Centreon connectors timeout¶
The timeout applied on checks execute by connectors were shorter than the timeout passed as parameter (near of 1s of difference). To fix this, timeouts are checked as milliseconds values now (more accurate) and the good duration is considered by connectors.
Volatile services and notifications¶
This version of Engine fixes the following behabiour: if a service is volatile, configured to send normal notifications. It won’t send anymore them if it is in downtime.
Centreon Engine 19.10.12¶
Multiline service check outputs were badly parsed and this lead to corruptions of status.dat/retention.dat files.
One dontime could be removed during the removal of an other downtime.
XXX_SERVICEGROUP_XXX external commands¶
XXX_SERVICEGROUP_XXX external commands were buggy. They take one servicegroup name as argument, it was resolved as a service name and not as a service group name.
Centreon Engine 19.10.11¶
Downtime cancellation was buggy¶
The downtime cancellation is fixed now.
Incoherence between hard and soft state types¶
A service could be in a hard state and the duration was not filled. Sometimes, it could also be the reverse. This is fixed and now displayed informations about soft/hard should be coherent.
TOTALHOST* TOTALSERVICES macros¶
Some global macros like TOTALHOST* and TOTALSERVICES were replaced by empty strings instead of numeric values.
Centreon Engine 19.10.10¶
Centreon Engine 19.10.9¶
Notifications on volatile services¶
On a volatile service, if notifications are disabled, it should not send notifications.
Recovery notifications must be sent even after an acknowledgement¶
With the new notifications implementation, this behaviour had been changed. This new version reset this functionality as users expect it to work.
External Command CHANGE_CUSTOM_X_VAR was not working¶
A bug was introduced on 19.10 for the extcmd CHANGE_CUSTOM_X_VAR. When we were updating a custom variable with a command, the updated value was always empty string.
Centreon Engine 19.10.8¶
Notifications and contiguous fixed downtimes¶
Between two contiguous downtimes, it was possible to receive notifications of devices that were in critical state whereas they had to be in downtime.
Second notification with state change¶
When a notification has been sent for a warning state. If the device changes to a critical state, no notification was sent. This is fixed in this version.
First notification delay if no recovery notification configured¶
When a first notification delay but no recovery notification are configured, the first notification delay is not applied. This is fixed in this release.
Double recovery notification sent on passive services/hosts¶
When a service come back to OK and is configured to send recovery notifications. It sends two of them instead of one.
Centreon Engine 19.10.7¶
Centreon Engine 19.10.6¶
Centreon Engine 19.10.5¶
Centreon Engine 19.10.4¶
Custom variable update crash on reload¶
A bug was introduce in 19.10.2 with a reload on custom_variables services/contacts/hosts update.
Recovery notification didn’t work on escalation [1;0]¶
If an escalation was defined with a first notification at 1 and a last notification at 0, we want the escalation to start from the first notification, but only one time, the recovery notification was not sent.
Centreon Engine 19.10.3¶
Centreon Engine 19.10.2¶
Custom variables not resolved¶
Some custom variables were not resolved. This results in bad or pending checks with wrong commands.
Random host statuses¶
Some hosts/services were changing status randomly.
Send custom variables to broker¶
Some broker notifications about customvariables were dropped.
Null string crash¶
In some cases strings construction could lead to exceptions thrown.
Centreon Engine 19.10.1¶
Centreon Engine 19.10¶
The notifications mechanism has been totally rewritten. First notification delay and last recovery notification delay should work as expected.
Pipes in results¶
Pipes are allowed in the centreon-engine plugin return value.
Migration of the code to C++11. This implies many changes in data structures. Some parts have also been rewritten and optimized.
The build directory is gone away. CMake is used as intended, this solves issues with some ide (like kdevelop)...
Better test coverage¶
We now have 123 new unit test (+63%). It allow us to have a better code coverage of the code base.