-----Original Message----- From: Kip, D. - GDI/SNB [mailto:d.kip at gdi.minjus.nl] Sent: Monday, March 09, 2009 9:52 AM To: hobbit at hswn.dk Subject: RE: [hobbit] Hobbit colors/icons
A recently red state should attract my attention, as well as a recent yellow state. But a longterm red state should practically be screaming at me and be as annoying as possible! Also a longterm yellow state should encourage me to make the problem go away. So all of these should at least have some form of blinking or movement to catch your eye, maybe slow breathing for the first 24 hours, then fast and at irregular intervals when they stay for too long.
[Cleaver, J.C.] Perhaps we're thinking too discretely on this. How about an SVG-generated (or pick of many pre-generated images) on a continuum of noticability? For example, a recently OK service could be based around a bright green -say, #00FF00- which slowly changes over the course of the next 24-48 hours to a #339933 icon (or whatever the background color is). Instead of simply "important" and "static" icons, let the display generator gradually modify the annoyance of a given problem.
Purple I have my doubts about. Of course it's bad when a check does
not
function properly. But at least to me, it basically only means someone forgot to start the hobbit/xymon client ;) I personally don't need much blinking annoyance here, but I may be missing out on something.
[Cleaver, J.C.] I agree that that's usually the only problem, but I think it's definitely something that should be given annoyance. In my experience, a "dead client" is often just waiting on something; caused by either a disk full condition, a hung NFS mount, or something else I do actually care about. But even if it died for no reason, that's still a "blind spot" in your monitoring and is IMHO dangerous.
-jc