Enable basic reporting for more system states

[sney]

Such as loadavg, and disk or service errors.

A low effort way to achieve this could be to hook nagios-style check scripts.

[mrdaemon]

I'm still mulling the "correct" way to implement this in an extensible, multiplatform way.
I'd like to at least have disk/raid/zfs status as a piece of data that can be queried ad-hoc, as it's something I'd like myself. I have aggregates of these in other monitoring tools, but "are all the disks/filesystems ok this instant" is a recurring question I have about my infrastructure.

"is there enough free space to apply patches" is a related one, which would somewhat justify its presence further.

Service monitoring feels a bit out of scope, in terms of "there are better written tools than my silly thing for this", but I am not really taking a hardline stance against it.

I think ultimately this should be a separate screen in the TUI, and a separate command with subcommands in the CLI. Open questions are still where to lob each platform specific implementation, and how to tie this into discovery.

This is not a high priority thing, but I'd like to experiment with it.

[sney]

"are all the disks/filesystems ok this instant" is a recurring question I have about my infrastructure.

Right, this is what I was thinking too. Just a different colour in the dashboard with the text "disk error" that I can then go and investigate through some other means.

Here's the nagios check_zfs plugin that I use: https://github.com/zlacelle/nagios_check_zfs_linux/blob/master/check_zfs.py
It takes a pool name as an argument and returns an exit code based on some OK/Warn/Error statuses. I've seen similar plugins for mdraid and even some hardware raid boards.

Service monitoring feels a bit out of scope

I wasn't thinking detailed service monitoring, but more like the disk status thing above. On modern linux this is probably just a matter of running "systemctl status --failed" and saying Error if it prints any output.

[mrdaemon]

Converting this to discussion, since adding this kind of functionality is a whole new featureset and requires thought

[mrdaemon]

I'd like to maybe add "needs restart" as extra status on hosts (see: #228) and this touches on the kind of foundational brick that may or may not pave the way for something like this, so it should be taken into consideration.