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
{{ message }}
This repository has been archived by the owner on Jan 15, 2019. It is now read-only.
Created by mfriedrich on 2012-01-04 10:04:10 +00:00
Assignee: mfriedrich
Status: Closed (closed on 2012-05-05 12:05:53 +00:00)
Target Version: (none)
Last Update: 2012-05-05 12:05:53 +00:00 (in Redmine)
protecting the check result lists in memory with a mutex ourselves we can make sure that modules could use the same mutex to gain exclusive access to that list.
at least DNX and mod_gearman use a special merge on that list before the reaping event happens, but this should be controlled by the core itsself.
especially when icinga-mq allows to pass check results via api, this will become mandatory to protect the core internal check result list.
found this elsewhere as well. seems to also affect the inner parts of the core itsself, not locking shared ressources properly.
add_check_result_to_list() and read_check_result() need mutex locking to prevent corruption of check_result_list.
Both the main and worker threads make calls to add_check_result_to_list but only the main thread calls read_check_result. However, both should acquire mutex locks for their modifications to the shared resource: check_result_list.
Updated by mfriedrich on 2012-02-18 14:33:49 +00:00
must be sort of an external addon in the above bug report, as this does not happen by default.
either way, such a lock would be nice to have when we implement -mq.
This issue has been migrated from Redmine: https://dev.icinga.com/issues/2242
Created by mfriedrich on 2012-01-04 10:04:10 +00:00
Assignee: mfriedrich
Status: Closed (closed on 2012-05-05 12:05:53 +00:00)
Target Version: (none)
Last Update: 2012-05-05 12:05:53 +00:00 (in Redmine)
protecting the check result lists in memory with a mutex ourselves we can make sure that modules could use the same mutex to gain exclusive access to that list.
at least DNX and mod_gearman use a special merge on that list before the reaping event happens, but this should be controlled by the core itsself.
especially when icinga-mq allows to pass check results via api, this will become mandatory to protect the core internal check result list.
Attachments
The text was updated successfully, but these errors were encountered: