I have seen this on a couple times on RHEL7 and Win2012R2 clients (v9.5.13) now where the clients are successfully reporting into BigFix and they show fixlets1, fixlet2, and fixlet3 as relevant. We have automation push a multiple action group containing the 3 fixlets to the target endpoint by the creation of an action via the bigfix REST API. The clients then evaluate the action and report back each fixlet as not relevant (The Fixlet which this action addresses is not relevant on this machine.) However, the Console and REST API still shows the fixlets as applicable on the clients. Why would the client report back as content is relevant but evaluate it as not relevant at time of execution? This happens for fixlets and tasks.
I can only think of two possibilities. The first is that the fixlets are actually not relevant, but the computer has not evaluated and reported back that they are not relevant. The second is that there is an issue building the multiple-action-group in your automation.
To check whether the fixlets are actually already not-relevant, it would be useful to use the fixlet debugger on the endpoint (or the WebUI Query app remotely) to evaluate each of the relevance clauses of one of those fixlets and make sure they all return True.
It’s possible that the fixlets are already not-relevant, but that hasn’t been reported back yet; the Action, on the other hand, gets evaluated/reported more immediately because it is “new content” and has a higher priority.
Are these out-of-box fixlets I could look at on my side, or are they custom fixlets written by your operators? There are some possible edge cases as well if they are using inspectors like ‘action’ that may give different results based on whether it is evaluating in ‘fixlet’ context or in ‘action’ context, but that would be pretty rare.
So our automation to build the MAG does work on 10’s of thousands of clients a month so it’s strange to see a few errors like this occur. When it happens it is all fixlets/tasks in an MAG and those fixlets/tasks can be custom made or external
For example in an occurance today, “BES Client Setting: CPU Usage” fixlet shows up as not relevant in the MAG but that fixlet is is and always should be relevant for all endpoints.