19:53 <brocifer> you got much experience w/ snmp checks in 1.2.0p2 man? Just
added a bunch of extreme switches and getting a ton of
annoying alerts related to wrong speed and some in/out-errors.
20:04 <brocifer> found a couple hacks, but it's just kind of annoying. thought
the wrong speed issue was fixed quite awhile back.
There is no wrong speed issue, and no link state issue - Those checks were initially made for systems where a link speed error IS a critical issue.
The issue is that you haven't yet configured them
21:04 <darkfaded> hey
21:04 <darkfaded> generally yes
21:05 <darkfaded> but the super elegant way i used to do it has become trickier
21:05 <darkfaded> so we should settle for "works, not elegant"
21:32 <darkfaded> you need to make a catchall rule
21:33 <darkfaded> this rule ignores link state, speed and errors for everything
21:33 <darkfaded> then you make rules that cover only your core network ports /
21:33 <darkfaded> those monitor for all of this
21:34 <darkfaded> but they also should have a reasonable level for error rates
(5warn/50crit) and the need averaging for 60 min
21:34 <darkfaded> this still doesn't cut it since the averaging code is a hack
21:35 <darkfaded> then you make a rule that gives those if checks 5
check_attempts or so
21:35 <darkfaded> chose at your discretion
21:35 <darkfaded> then also make an exception for your servers
21:35 <darkfaded> there, i'd go with 20W/40C
21:35 <darkfaded> same retries
21:36 <darkfaded> this should sort it
21:36 <darkfaded> if you can't identify your uplink ports via their name or
their port number, you need to find another way
21:36 <darkfaded> that is for example, to only monitor the port states / error
rates on the core switches
21:37 <darkfaded> or to disable alerts for all edge switches (you still get the
alert from the core switch if a uplink fails (50% chance with
fibre lol) and you'll also of course see if the edge switch
goes host down
21:38 <darkfaded> you cannot, without tricks achieve to have something more
powerful than that
Remember the if speed and state are saved on inventory if configured to the defaults.
Some of these changes need a reinventory. If that doesn't help, use cmk -D to debug if the expected rules / levels are in effect, or even better yet, hit the rule match viewer from WATO's inventory menu.
Advanced config magick
You can also work by disabling "Warning" notifications for certain service groups
You can also make your Dashboard more helpful by copying the default views, making a replacement dashboard using those and then only displaying HARD states. This will remove the jitter.
my old blog post about the inner workings of ignoring stuff.
(If you don't need stats, and don't need alerts, why are you even inventorizing your desktop's uplink port in the first place!)