Looking for the right monitoring software

This forum was archived to /woltlab and is now in read-only mode.
  • Hello,
    We are a small IT company that supplies mission critical software to our clients. The software is a low budget passenger management system for cab drivers and is targeted at small home businesses.
    It is therefore installed on a mixture of W2003, W2012 and more lately W10 machines.
    We have for many years been plagued with avoidable out of hours support as a result of disk full, internet off, backup failed and other avoidable situations that may occur in our own application.
    I am looking for monitoring software that will visually alert us to any “triggers” that we configure in the agent. This will enable us to reduce the unscheduled maintenance so we can be proactive rather than reactive to imminent failures.


    Is Icinga the correct monitoring tool for me?
    If so where can I get some assistance in the basics for setup and distribution of agents?


    Regards
    Stewart

  • Hi,


    I'd say you are best with Icinga 2 being the core check execution and alert engine (and agent nodes) and Icinga Web 2 providing visualization on your monitoring data be it alerts, details, etc. In additiona to that there are several addon integrations available for example graphs, maps, ticket systems, business processes, etc.


    Pick your preferred distribution (e.g. CentOS 7) and start with the documentation for Icinga 2 on docs.icinga.org. It will guide you on how to install Icinga 2, the IDO database, Icinga Web 2 from packages. Once completed you'll step further into the configuration details, and of course add your Windows clients. One thing you should keep in mind when using the Icinga 2 client - W2003 is EOL and therefore not supported. Extra-addons such as NSClient++ will help out, but you probably should consider an upgrade to a modern Windows before proceeding.


    In case you are looking for a graphical interface for configuration, you might also give the new Icinga Director a try.


    If you are wondering how Icinga 2 and Icinga Web 2 look like, and just give it a whirl inside a demo environment, please use the available Vagrant boxes to play with. Note: Icinga 1.x is the old world inherited from Nagios as a fork. Icinga 2* is rewritten from scratch and our main development target for the next decade.


    If you happen to have specific questions or problems with your setup, just chime in here, open a new thread, provide all the details (distribution, versions, configs) and we are trying our best to help you get things going :)


    Kind regards,
    Michael

  • Hi Michael,
    Thanks for the reply.
    I hear you about the EOL products, however we are offering IT support to third party installations so we cannot always dictate the OS.
    As I mentioned in some cases the machine we need to monitor may be a free standing W10 device.
    I have checked the docs and although it all makes informative reading I still do not get a clear picture as to how we can bundle the agent to deploy with our software so that each new user that installs our software will automatically reflect on the portal to enable us monitor the critical parameters.


    Regards
    Stewart

  • I'm not sure whether the Icinga 2 client satisfies your needs here with bundled installers. One thing which would probably cause trouble is the windows setup wizard with csr autosigning (that should probably happen invisible to the user). I guess some sort of silent install, and then manually prepp'ing the configuration and ssl certificates for the client (maybe injecting them before windows is finally booting up) could work, but I've never tried such sort of bundling. Maybe it is easier to use the (insecure) nrpe with nsclient++ approach here.

  • Sorry you got me with the Jargon "it is easier to use the (insecure) nrpe with nsclient++ approach here."
    I dont understand any of the acronyms, I tried searching on the internet but it did not help much
    Are you now talking about another product or is this still Icinga?


    Stewart

  • That's an addon called NSClient++. It offers various client cli plugins, the most common one is check_nrpe or check_nt. Both use a somewhat insecure encryption layer.


    https://www.nsclient.org


    Icinga 2 already provides CheckCommand templates for check_nrpe, check_nt.

  • Thanks for the info..
    I have been taking a good look at NSclient and initially it appeared to do what I wanted. It looks like the setup is a bet tedious but I am sure we can script all of that so it can happen unatended.
    What I cannot seem to get a full understanding of is the communications architecture.
    It looks to me like the monitoring server needs to be able to Reach out to the agent to collect the data.
    If this is the case then it is a show stopper as we have no control over the client routers and could not forward the ports.
    Is my assuimption correct?
    If so I will need an agent that makes an outbouind connection to a server and not requiring an inbound connection from the monitoring software.
    If I have misunderstood could you clarify the registration and communication process between the agent and the hosted Icinga installation?


    Regards
    Stewart

  • NSClient++ provides different communication protocols. The most common one is to query the service from the monitoring master via check_nrpe. It could als send checks as passive results to the core using send_nsca.


    Both ways (nrpe, nsca) worked for many years, but have flaws with security implementations. Given that your satellites/clients will communicate over the internet and not some local (v)lan, I'm not sure if you should go that route.


    As mentioned before, that could be a solution for 2k3, as this is in the same EOL unsupported state like XP.


    If you are looking for a modern alternative, the Icinga 2 nodes allow connections from both ends, be it the master connecting to the client, or the client to the master. The preferred configuration way here is the top-down cluster config sync. I.e. if the client is deployed the master adds the configuration (default set of checks, etc) and pushes that to the client/satellite. The client will run standalone and execute checks. On connection loss, it will store its own local replay log in order to pass collected check results later to the master (sometimes relevant for SLA reports). Both communication ways work, there will only be one active network connection between the two nodes.

  • It sounds like the Incinga 2 Nodes will be ideal for us. Will they work on W10 as well as 2003 and 2012?
    Also I will need some guidance and support setting up Incinga2 on our Amazon servers and then configuring a client. This is something we would be prpared to pay for. Who would be best to approach?


    Regards
    Stewart

  • As said, 2k3 is EOL and might require a workaround. Though that could be solved via nsclient as addon only for those "old" systems.


    In terms of professional support or consultancy - we are currently building an Icinga partner network. My employer is one of them (obviously :)), you can find them over here: https://www.icinga.org/partners/

  • I have tried the one in the UK and its a mobile phone on voicemail.
    Please direct me to a commercial partner who will be able to assist


    Regards
    Stewart

  • Hi Thanks for the reply.
    We seem to be managing with the hosting on Amazon.
    We have setup Incinga 2 web and I can login and navigate ok.


    I would now like to try and get some reporting from one of our servers.
    What is the most simple way I can deploy an agent that will make an outbound connection to the web hosted software.
    Is there any quick start guide?


    Regards
    Stewart