I'm in the process of changing a bunch of our monitoring stuff, and smartctl as usual is problematic.
Most of the time, looking for "No Errors Logged" in
smartctl -l error <options> seems to do the trick, but for example some disks with ECC have a very different output that won't include that even when everything is fine.
Is there a guide somewhere of what should and shouldn't be considered a problem in smart's output for each disk type ? For example, for those ECC disks, it seems like I'd need to actually parse the output and figure out if the number of correction is going up, which seems overly complicated (especially since smartctl still doesn't have a json output format :( one day). I can't just see if the self tests are working, because apparently the self test can have PASSED status even if the disk logged errors, so that has to be a separate test.
How do you all monitor your smarts ?