I would like to propose the following new trigger conditions.
1. Condition based on node (self explanatory i think)
2. Condition based on 'application'. Either the application of the item(s) (preferred i think) used by the trigger or adding the application context to the triggers.
Why:
Right now most of the conditions are based on 'hosts' or trigger constants, which means that all problems on a host can trigger a message or action. What i am missing in zabbix is the ability to base actions (specifically for sending messages) on a certain subject like 'hardware', 'oracle', 'operating system', 'network', etc.
There are ways to achieve similar functionality right now, but these methods either lack the ability to use layered templates or require all trigger messages to have 'match strings' which is ugly and puts unnecessary strain on the database (wildcard queries).
Example:
Large environment, layered templates, multiple support groups who work independently of each other (think multiple stand-by schedules, NOC, 3rd party specialists, etc).
Server is running an oracle database. Zabbix is checking hardware, OS, oracle and several smaller processes. Therefor the server is placed in an OS specific hostgroup and the Oracle hostgroup, users have access to these hostgroups.
Now this server's OS is maintained by a group of sysadmins, oracle is maintained by DBA's, etc. So in the current situation both user groups would receive all triggers, because both groups are related to this 'server'.
I would like to be able to only send oracle related issues to the oracle support group, and OS related stuff to the OS group.
Furthermore, the 'application' system allows for items to have multiple applications related to them, making it easier to catch those 'grey areas'.
Comments please
- Are there more people seeing a use for these 2 proposed action conditions?
- Developers: if something like this feasible (easy enough to add to 1.6 or hard enough to move it to another version)?
1. Condition based on node (self explanatory i think)
2. Condition based on 'application'. Either the application of the item(s) (preferred i think) used by the trigger or adding the application context to the triggers.
Why:
Right now most of the conditions are based on 'hosts' or trigger constants, which means that all problems on a host can trigger a message or action. What i am missing in zabbix is the ability to base actions (specifically for sending messages) on a certain subject like 'hardware', 'oracle', 'operating system', 'network', etc.
There are ways to achieve similar functionality right now, but these methods either lack the ability to use layered templates or require all trigger messages to have 'match strings' which is ugly and puts unnecessary strain on the database (wildcard queries).
Example:
Large environment, layered templates, multiple support groups who work independently of each other (think multiple stand-by schedules, NOC, 3rd party specialists, etc).
Server is running an oracle database. Zabbix is checking hardware, OS, oracle and several smaller processes. Therefor the server is placed in an OS specific hostgroup and the Oracle hostgroup, users have access to these hostgroups.
Now this server's OS is maintained by a group of sysadmins, oracle is maintained by DBA's, etc. So in the current situation both user groups would receive all triggers, because both groups are related to this 'server'.
I would like to be able to only send oracle related issues to the oracle support group, and OS related stuff to the OS group.
Furthermore, the 'application' system allows for items to have multiple applications related to them, making it easier to catch those 'grey areas'.
Comments please
- Are there more people seeing a use for these 2 proposed action conditions?
- Developers: if something like this feasible (easy enough to add to 1.6 or hard enough to move it to another version)?
Comment