"top", "htop", "nmon", etc. are great for monitoring processes on machines.

Top-down, imagine them as frameworks for displaying, and adjusting the display of, structured data whose input reads data associated with UNIX processes.

Now, we have an application whose monitoring could greatly benefit from this sort of structure, except that the input isn't processes but other changing data.

As far as I know, none of those three programs have the ability to monitor data from other sources. (Correct me if I'm wrong on that!)

Is there a similar monitoring program with (for instance) the capability for the end-user to provide a local plugin (or similar) for their local data?

Only a process can create "changing data". Data cannot change on its own (barring cosmic rays or other device failures).

Now for monitoring FILES that may be altered you can look into the inotify tools.

https://github.com/rvoicilas/inotify-tools/wiki


I know of no tools like top for user data. Top monitors only the kernel process data tree, and only the kernel updates that data. For any other "data monitoring" you would require an interface into each of the applications that modify data... And I don't think that is what you are thinking abut.

Well, let's sidestep the interface for now. What "other changing data" do you mean? Care to give some specific examples?

Thanks for the reply. Appreciated! I should probably have added that our changing data is in shared memory (and there is also some from processes).

So I'm thinking of a top-like program. But rather than directly reading (or periodically sampling) process information, it could instead read from a user-supplied plugin that gathers data from shared memory or other sources, including processes. (So a top-like program with a level of decoupling between the user-oriented display/control, and a data-gathering plugin. The default plugin would (on a Linux) system look in "/proc"; a user could supply their own alternative plugin.)

For some reason the phrase "specific examples" doesn't seem to be parsed here as "give me exact details I can work with" so that excludes me from contributing further.

Good luck!

Nxlog, rsyslog, custom bash scripts,.. send data to graylog or elasticsearch. IIUC this is a fairly normal procedure. Nxlog will read a file, parse its contents and spit it into elasticsearch. Rsyslog can do the same thing. Nxlog and Rsyslog are log gatherers/data gatherers, and elasticsearch is the repository where that data is stored.