Forum Discussion

JamesCC's avatar
JamesCC
Icon for Contributor III rankContributor III
4 days ago

Understanding DA caching in a CUG

We have a situation where a single DA aligned with a root device caches data about component devices which it then uses to populate metrics for both the root and component devices.

Does anyone have some more detail on what happens - at a CUG level - if the collector aligned to the root device becomes unavailable? How does another collector in the CUG know that it needs to start caching data (is that via config push?) and does the whole DCM tree need to be rebuilt but aligned with a different collector?

Equally - is there any smarter way to avoid the "Device failed availability" event for the child devices which is only occurring because the cached data from the root device is not available (and not because the child devices really have an availability problem)>

device_management

2 Replies

Sort By
  • ErickBurgess's avatar
    ErickBurgess
    Icon for Moderator rankModerator
    3 days ago

    Make sure you have the "Collector Affinity" setting correct on all the DAs in this PowerPack.  The DAs aligned to component devices that read from a cache should be set to "Root device collector."  If you have any DAs aligned to a component devices that don't read from a cache, then set them to "Assigned collector."  The setting on the DAs aligned to the root device should be "Assigned collector."  Setting it to "Root device collector" wouldn't cause a problem because it is the root device.  As long as Collector Affinity is set correctly, it doesn't matter which collector in the CUG the component devices are aligned.

    Regarding Component availability, one option would be to code the logic in the Snippet to return Available unless the result read from the cache specifically indicate the device is not available.  Meaning if the cache isn't accessible, the Availability Collection Object still provides Available as the result.  If the method used to collect the cached data isn't working (API didn't respond), the Availability Collection Object still provides Available as the result.

    Regards,

    Erick

  • Issac's avatar
    Issac
    Icon for Expert rankExpert
    4 days ago

    As far i know when DC is down in a CUG ,the devices which are present in that DC will be rebalanced to the available ones and the config push sends updates about the new devices registered to CDB.  Sometime due to time taken by config push device down alerts may trigger . Even we encountered this . May be we can tweak the Time over Threshold in event policy to avoid this .