I would like an answer to this too.
thank you for your feedback and I am sorry for my late answer. What you describe was an inital behavior of NPM RC1-3 and has been reported several times by customers (see e.g. Re: NPM 10.4 RC1 Hardware health on Juniper Switches - Virtual Chassis or Re: Hardware Monitoring Issues with Juniper Devices).
Based on the customer feedback, we have decided to change this behavior and starting with RC4 standby status is reported as Up with the green icon.
Hope this answers your question.
just to answer part of yours original question. The meaning of health state of polled sensors is not configurable.
So a new "feature" was again turned on, not listed in the enhancements, and we can't configure it.
Thanks SW, now I have a log full of warnings that I have to explain to our Help Desk who feels the need to alert us every time they see something in warning.
No, I don't want to turn off the logging, because it was actually helpful. However, to say that every single port on my 6509's is in warning due to the fact that there is a VOIP phone plugged in, is horrible programing. These are POE ports and the current load is expected.
As I stated to your tech support RE the horrific 10.4 installer, the QA on this version is lacking.
I'd like to see a feature request to allow the threshold for health components to be editable.
I understand your problem and we are trying to collect all similar feedback like this to make the right decision for the future. Unfortunately our QA was not able to test all existing devices. We already have some feedback from RC phase and I may say that we are working on solution how to solve most painful problem you guys experiencing regarding HW health. Please stay tuned for more.
I'm sorry, but I would argue that the 650X and 3750 Cisco series is one of the most popular versions of hardware, and any testing on those platforms would have resulted in the same findings.
The real ball drop here, is the continued lack of documentation from SW when it comes to features which have been added. In addition, the techincal support staff is not trained in the new verions, and learn the fixes in my via PROD environment.
It unfortunate too because the new tools have great value to add, however with the instant bad taste, my senior engineers and the rest of my Ops staff are done with this enhancement. I've removed its ability to log anything to our events.
Sadly I knew posts like this were coming and as such I have held off upgrading to 10.4 even though it fixes some outstanding issues. I run 3750 switches too so will hold off until 10.x but fear the problem will just be in another area.
Can we have a poll for new features that includes "greater QA testing" as an option? I suspect it may end up being the top selected feature request.