mnagel

Members
  • Content Count

    494
  • Joined

  • Last visited

  • Days Won

    87

Posts posted by mnagel

  1. 22 hours ago, mnagel said:

    Thank you!  I have been asking for this via "proper" channels for some time with no results -- will try it out ASAP as I have an 8320 cluster waiting.

    FWIW, I recommend using a standard property name alongside the ssh.user/ssh.pass (e.g., lmconfig.enabled) to allow disabling this premium feature at the group (client) level when it has not been subscribed to.  I know it is an uphill battle to get those all fixed, but I sure wish it could be done.  We still cannot use the new AD and DHCP modules due to lack of ability to disable LMConfig per client.

    I guess not ASAP:

    This LogicModule is currently undergoing security review. It will be available for import only after our engineers have validated the scripted elements.

    I guess I will check back at some indeterminate date in the future :(.

  2. Thank you!  I have been asking for this via "proper" channels for some time with no results -- will try it out ASAP as I have an 8320 cluster waiting.

    FWIW, I recommend using a standard property name alongside the ssh.user/ssh.pass (e.g., lmconfig.enabled) to allow disabling this premium feature at the group (client) level when it has not been subscribed to.  I know it is an uphill battle to get those all fixed, but I sure wish it could be done.  We still cannot use the new AD and DHCP modules due to lack of ability to disable LMConfig per client.

  3. 20 minutes ago, Stuart Weenig said:

    The author needs to verify that the module has been published to the public repository. In simple cases, it's automatically made public at that point. If there is code in the module, it will undergo manual security review by the LM staff before it is made publicly available.

    Almost certainly there is code as Palo Alto checks virtually always require API access.  Review has seemed in most cases I have been involved with to be a mostly ad hoc process (or if not, definitely opaque). I suggested in one of our UI/UX meetings that there be a "Request Review" button or similar to create or escalate a request for security review.  As a bonus, use a ticketing system (this would be welcome for feedback as well, which as I understand generates internal-only tickets).  A unified customer visible ticket system for feedback and module review would be very helpful.

    • Like 1
  4. Been there, done that -- you can't reference those in widgets, sadly.  You have to just create your own datasource that sets values equal to the properties and then reference those.  First time I ran into this I wanted to chart device usage against subscription levels, latter of which was a property.  In ours, the collector is a Groovy script that does nothing (not sure why that was how we did it, but it works).  The CDP is just equal to ##property## in each case.  It is Groovy mode, but the code is literally just that.

    image.png.7b827a81fc6ce526b78f5882e504eea8.png

    • Like 2
    • Thanks 1
  5. FWIW, having also come originally from Nagios, I miss the ability to transmit arbitrary string data back via alerts. Some of this can be emulated with auto properties, but those can be set only during discovery not collection.  I posted a feature request previously to allow definition of enums that can be bound to datapoints (global values and overridden values within specific datasources/datapoints). these could then be used to avoid the current awkward legend method and actually show the intended purpose of DP values where needed via tokens. Imagine a line that showed the actual meaning of the current value instead of a long truncated legend line that makes you dig around for what it means.

    I also think it should be possible to improve the property menus to leverage more advanced typing and UI.  For example, a property might be just a string as now (preferably with better input box control), or it might be a radio button, selection menu, etc. so that folks using properties can easily find what is supported and what values/ranges are allowed.  This also would be something where those hints would be defined within logicmodules primarily, but it should be possible to define them more generally (at least the typing/UI definitions, which could then be bound to properties that are used within modules.  This is not strictly related to the topic, but is about readability and usability so I tossed it in there, too :).

    • Like 1
  6. On 3/6/2020 at 6:35 AM, Stuart Weenig said:

    You would then need to look into changing strategies over to an EventSource. The EventSource would output words (like a log file) and you'd write a check to look for particular words to open a specific kind of alarm.  

    Eventsources don't support embedded Powershell, though they certainly should.  You can upload a script though.  That said, eventsources are also almost entirely unsuited for monitoring, more like additional information to see along with monitoring. Among other things, you cannot ACK them in a meaningful way due to lack of correlation across eventsource results.  I'm sure the yet-another-premium-module LMLogs will fix all those problems, though.

  7. I would not hold your breath -- I have had to fight just to get and keep SPF enabled on our email.  Regardless, even if you could use the builtin alerts with a distinct From address, you would still have portal links embedded in the message that reveal it is LogicMonitor.  You could do what we do and submit everything via a custom email integration (or a web integration via an API handler), then handle the data any way you like.  In our case, we feed the tokens into an actual template system to format messages using conditional logic and all that stuff missing in the LM blind token substitution method available normally.  We feed that transformed result into a ticket, but obviously it could be handled many different ways at that point, including re-routing via email with proper headers.  The downside of a custom integration is that there is a bug -- certain things LM simply does not send to those that are sent with the builtin integration (e.g., ACK and SDT notices). I have asked about this and in theory it might be fixed one day, but it has not been in well over a year since reported.

    • Upvote 2
  8. Here are at least two items that need to be added to make the dashboard token feature more useful:

    • adjust widgets that cannot use tokens so they can (e.g., Alerts, Netflow, etc.)
    • allow arbitrary tokens to be inserted as needed within widget fields (e.g., device patterns, instance patterns, etc.)

    A concrete example of the latter came upon me this morning.  We have multiple locations with similar equipment for which we want to display Internet usage details, one set per dashboard (cgraph and netflow widgets).  The edge device names vary as do the uplink ports to the ISPs in each location.  Cloning this dashboard solves virtually nothing as every single widget still requires editing.  If the tokens could be used, these dashboards could be cloned without the manual editing other than filling in the necessary tokens.  In some cases the tokens are insertable, but most fields do not allow them.  In this case, I defined various tokens like isp_1_name, isp_1_edge_device, isp_1_edge_port, etc. but could use them in very limited ways ultimately making the exercise pointless.

    As with many things, we can at least workaround this with the API (at least I believe I could with some effort), but it would be much more accessible to folks if handled within the UI.

    • Like 2
  9. 5 minutes ago, mnagel said:

    The normal way I monitor services is via AD, but you would end up with a new instance wildvalue each time it was changed if you use the normal option (WMI-based datasource).  If you use Groovy script DS instead, you could strip the PID portion to build the wildvalue so that the data is stable.  There should be some examples of that in the existing datasource repo, need to dig around....

    Many examples of using WMI from Groovy, none that select from Win32_Service, but should be simple enough to adjust the query.  See Microsoft_LyncServer_StorageService as one example.

    • Like 1
    • Upvote 1
  10. 2 minutes ago, Stuart Weenig said:

    AppliesTo Functions is a misnomer. They should be AppliesTo Aliases.

    Yeah, brings back horrible memories of me requesting repeatedly the documentation on how to pass parameters and getting the most insane response from support :).

  11. 6 minutes ago, dcyriac said:


    Wish for a feature to capture the LogicMonitor settings and state of all the resources, resource credentials, websites, mapping, reports, exchanges and settings.

    Perhaps an offline back so that we can restore monitoring for environments that were deleted without having to go through the scans or process of adding devices individually.

    Or to revert LogicMonitor to a state before or after an account impacting change.

    Best we have been able to do here is a script leveraging the API to download as many endpoints as we are able to access with checkin to a git repo.  Works, but needs frequent tweaking as things change on the backend. Having a way to revert to a previous snapshot or similar would be very handy.  My script came about originally after I implemented alert rule resequencing with an error and lost some rules.  My latest incarnation of this script has an option to check the items as well for problems (e.g., broken widgets).

  12. Since I can't wait for this, we now have code to grab widget data for all supported widget types incorporated into our existing backup script (pulls virtually anything I can from the API into a Git repo regularly).  Most issues can be detected via exception (non-200 status code), some require a bit more analysis (no data in any line in a cgraph, for example). Working reasonably well now for the first phase, which is to be aware of busted widgets before we are embarrassed during client review. Next phase will be to analyze data more specifically to the context (once I figure out how to represent the widget check requirements).

  13. I am incorporating this into my resource check script.  For now, the first item I am testing is Windows_DHCP, which requires that the DHCP Server role is installed (we have an auto.winfeatures property that is populated by a PropertySource. I don't recall where we got it, somewhere in these forums IIRC :).  The PS code is simple:

    hostname=hostProps.get("system.hostname")
    my_query="Select NAME from Win32_serverfeature"
    def session = WMI.open(hostname);
    import com.santaba.agent.groovyapi.win32.WMI
    def result = session.queryAll("CIMv2", my_query, 15);
    println "WinFeatures=" + result.NAME

     

    If this list does not include DHCP Server and we have Windows_DHCP assigned, it will trigger a warning.  I plan to extend this to catch more stale categories.

    My check script tests a bunch of things, including lack of any FQDN or expected FQDNs, lack of NetFlow data (the new heartbeat datasource is not helpful there as it does not care if valid data arrives), and other stuff that can go wooorng.

  14. It seems like a bug to me that if you assign Thresholds in a role that you are allowed to edit thresholds, but not toggle alerts or notifications on and off.  Please either include that within Thresholds, or add another permission to enable that without having to assign full Manager permissions.

    • Upvote 1
  15. Unfortunately, they will need to make instance groups into actual groups first, which I have requested many times.  As it stands they are basically limited tags where you can only have a single tag for any instance and then you can apply thresholds to the tag, but those don't apply to new items tagged the same way.  This is why they throw up the warning about how new instances won't get thresholds automatically.  We've had to workaround that with API scripts to refresh those instance group tags and thresholds when new volumes are added to servers (as just one example). I assume if they ever are made into actual groups, properties would be part of that.

    • Upvote 3
  16. I would start here: https://openvpn.net/community-resources/management-interface/

    There is an example of how to use this (not in an LM-friendly way) at https://kifarunix.com/how-to-monitor-openvpn-connections-using-openvpn-monitor-tool/

    There is also an example (a bit dated) on how to expose data via SNMP here: https://github.com/Phhere/openvpn-snmp

    The problem seems to be a general lack of standard monitoring since OpenVPN runs on so many platforms.  If check_nrpe works, perhaps just punt and use that via an LM script datasource :).

    • Like 1
  17. It sounds like you are expecting a function to be able to take arguments, because "function".  I did as well many years ago when I wanted to create a single isClient function with the client name as an argument, but found after a very painful support ticket that they absolutely are not functions, just macros.

    You would need to write a separate version for each check.  In this case, you would have to add a new function like this and a new one for each version:

    system.virtualization=~"VMware" && auto.version_number =~"6\\.5\\.0*"

     

    • Like 2
  18. I would like to have SSL expiration profiles I can apply to websites with automatic selection based on the CA. This is (for now) based on the fact that Let's Encrypt certificates tend to have short expiration intervals, but it seems like a good general solution to discover stuff automatically as elsewhere within LM.

    Thanks,
    Mark

    • Upvote 1
  19. We have for some time used LucidChart to present network diagrams to our clients via an iframe in a widget. The Topology Mapping feature has its place, but it is not a substitute for diagramming. Among other things, TM has no support for most WAN topologies.

    I had not looked into it much before, but LucidChart has a Data Service API and in theory, various device details could be replicated via that method (https://www.lucidchart.com/pages/api-data-service).  It will take me some time to digest that and figure out how that would be useful for presenting information from LM, but it sure looks like it could do it.

    • Like 1