Before a restart of the application it would be great to save all states services and hosts, to avoid alert replication at restart for all the warnings and criticals events already there.
I've been keeping an eye on your product, and it looks incredible. However, due to institutional policy we cannot use the software if there is no enterprise support (which we would pay for, of course).
Create a poller module using the Python OpenOPC library to read VTQ (Value, time quality) tuples from an OPC server. The poller module uses threshold values from the Shinken service definition associated with the data point or tag name.
including more control on python processes for the services (not using srvany),
but using .NET services installations.
a XML conf file is already used for each service and for the moment a small UI is designed to change the configurations.
The installation will be done using InstallShield, each service as a feature so with the choice to install only one service... more »
HMAC message digests for signing pyro messages. (Pyro is the messageing framework that Shinken runs on). This feature is available starting at Pyro 4.04.
This would be a good first step in securing default Shinken installations.
Calling remote plugins is a mess (as in nagios); I'd like to see something like a "via" keyword to service configurations.
This would call a rpc-plugin, meta-plugin (call it as you like) which is used to call the remote plugin.
At least nrpe and ssh should be implemented.
plugindir should be a dir-list, maybe rename it to pluginpath
pluginpath = /usr/lib/shinken/plugins;/usr/lib/nagios/plugins
I would suggest to move the "Requirements" section in the README.rst file to the top, maybe right after the useradd section and before the "First way: ..." section. Would make more sense in my opinion.
The most common way of doing this, is to store to a file on disk, this is heavily I/O intensive when there's 1000's of checks a minute doing this.
Suggestion: Create an API to graphite, via a few extra paramters, either available externally or handled inside shinken, allow for... more »
Like the module for VMWare, it would be a good idea to generate the dependencies between hypervisors and virtual machines for Open Source solutions like Xen / KVM.
This kind of discovery should be real faster than nmap to track existing hosts.
Then, when could populate a field called "neighbours" in the hosts. The field would not be used the reachability function, but only in the maps. Using only parent-child relationship is not enough to draw a map of the network.
Using the js thejit toolkit, already used in the "dependancy"... more »
Il will be great to define a chat room where alerts could be sent by a shinken bot. A user can ask bot to have the list of alerts and, maybe, who is on.
A user can ACK an alert to the bot.
Adjust thresholds is just boring and can take times to adjust a new behavior that make obsolete old thresholds.
Why not having adaptive thresholds ? Via a directive, you specify a number of occurrences that make shinken takes the significant thresholds. Or just provides different level of adaptive choices : low, medium, high with predefined parameter.
Or takes metrics average and calculates its thresholds with a... more »