9/7/99 -- 3:23 PM, AP Breaking News
[ Fair Use: For Educational/Research Purposes Only]

Reports of temperatures, weather conditions delayed

WASHINGTON (AP) - A breakdown in National Weather Service computer hardware Tuesday morning was blamed for delays in reporting temperatures and weather conditions across the nation.

It was the first time in 20 years of service that the computer has had failures in both of two duplicate systems, said Art Wick, chief of the networking and communications branch at the National Center for Environmental Prediction.

Temperatures and other weather data are collected across the country by the system, called AFOS, which delivers the data to the center in suburban Camp Springs, Md. AFOS then relays tables and summaries of the data to users such as newspapers.

The data collection system was working, but the computer failure prevented AFOS from forwarding the information to outside users, Wick said. He said the system was back in operation by midafternoon.

``AFOS is configured so there are two sides to the system. Ordinarily we can operate fully with either side up. ... For the first time we had a hardware failure on both sides,'' he said.

AFOS stands for automation of field operations and services.
-------------------------------------------------------------------

-- Ashton & Leska (allaha@earthlink.net), September 07, 1999

Answers

Some of the programmers, technies, and systems experts can explain this?

-- curious (all@systems.down), September 07, 1999.

This is an example of what appears to be an older (mainframe) system where they had two "sides" to the system for redundency. The fact that they had the second side shows that they were aware of the vulnerability of the equipment to failure and had the second for "contingency" while they fixed the first one.

Having them both go down at the same time once in 20 years is pretty good.

Reminds me of the time I was running diagnostics using one side to check the other and the "other" had a bad bit so I finally figured out that I had to use ASCII and subtract the letters that should have been written feom the ones that were written to figure out (in binary) which bit was bad in the second one- fixed it and went on to use the diagnostics to fix the first problem.

I have to tell you I was sweating that one out until I came up with what to do about it! But then that is what troubleshooting is all about... not knowing what the problem is and going in and using troubleshhooting techniques to isolate the problem and then fixing it.

-- Cherri (sams@brigadoon.com), September 08, 1999.

Weather is one of those variables that could really complicate Y2K contingency planning in the middle of this winter. Not being able to accurately predict severe weather adds another nasty layer.

-- Brooks (brooksbie@hotmail.com), September 08, 1999.

And not off topic at all: did anyone see what I did on the DC weather report over ABC Monday evening: that they expect severe weather, mainly snowstorms, in much of the midwest and two other wide areas they marked on the weather chart, further west (missed locations). That doesn't bode well, does it?

-- Elaine Seavey (Gods1sheep@aol.com), September 08, 1999.

Well, we read a Farmer's Almanac article that predicted 4 major snowstorms Nov-Dec, very wet & cold, then mild after 1/1/2000 :-)
Sure hope the mild part is accurate ...

Cherri, thanks for your explanation, maybe that's what & all it was. Kinda suspicious o deez failures at this point!

The upcoming "3-day storm" promises to be one of the most unusual in history, to be long studied and argued about afterwards ;^)

-- Ashton & Leska in Cascadia (allaha@earthlink.net), September 08, 1999.