I know that "Flush All" solves those issues that display something wrong in the UI.
However, I don't see the need to flush caches for components that are working fine, so I would like to avoid the use of "Flush All" and flush just this/these cache/s which related with the symptom.
Furthermore I have seen that flushing many or All single caches (consecutively ) doesn't produce the same result than "Flush a Group" or "Flush All".
As a real example, not being able to display a section with a calculated attribute I flushed many caches and even including those in "Custom Objects" group (odfObjects, customViews, odfObjectMap, dataProviders, validationHandlers) without success.
However, I flushed the group "Custom Objects" and the issue was immediately solved, and even the Used Memory decreased in a higher percentage.
Is there any document covering the caches with technical details (specifying what means all the names of single caches and where can impact)?
Flush Caches is not a feature provided and you should never flush caches to over come an issue. If you are flushing an object then you are actually breaking the transaction ongoing by any other users. If you are seeing these incosistencies then its the problem with your multicast and that needs to be fixed.
But this method is used with success by many users and even recognized by CA as a tip/workaround
CA PPM Tuesday Tip: How to ensure data is displayed properly in the browser.
About these inconsistencies, in my opinion this is not an general issue because this display problem doesn't happen frequently (just few times in a year) and our clarity service is not a cluster.
From admin tower, monitoring using trace on, multicast messages are received..
Am i missing any?
If you see the below thread we also recommended not to use as its not a fix but masking of a problem which is not beneficial in long term. Also I would like to get your environment reviewed on the multicast by Support and we can suggest. Clearing caches is never a permanent solution.