Occasionally, the feed returns weather data from the beginning of January.

  • 10
  • Problem
  • Updated 4 days ago
I've been experiencing this for over a month.  I was hoping it was a temporary glitch.  Anyone else having the same issue?  Any fix?  Thanks.
Photo of Jerry Y

Jerry Y

  • 2 Posts
  • 0 Reply Likes

Posted 2 weeks ago

  • 10
Photo of dobry

dobry

  • 21 Posts
  • 1 Reply Like
We have same problem since a month and it became worse past days. It shows day forecast from January and no hourly forecast. We get several claims from users every day and can within short time reproduce the problem. 

It happens for example with such call:
http://api.wunderground.com/api/<MY-API-KEY>/conditions/astronomy/forecast10day/hourly10day/la...

As usual Wunderground support (help@wunderground.com) doesn't react at all. For what we pay $6'000 a year? For such underground support and data !?!  Upset!
Photo of Andrzej

Andrzej

  • 1 Post
  • 1 Reply Like
Same here, using wunderground for Kindle weather.

Few times a day (updating hourly) an Wednesday old data is downloaded. Always same day, conditions and forecast.
Photo of naltatis

naltatis

  • 1 Post
  • 0 Reply Likes
Same here. First noticed this effect an May 10th.
Photo of Stefan Cojocaru

Stefan Cojocaru

  • 8 Posts
  • 11 Reply Likes
Sent an email to them with this bug report.

Hope they fix it soon

Photo of Murphy

Murphy

  • 19 Posts
  • 10 Reply Likes
I have been seeing the same thing for about the same amount of time. Sometimes, the weather forecasts will be fine and then they revert to a wintery forecast for hours or days. They they change back to accurate.  It's infuriating.  Weather Underground is one of the most expensive parts of our system, and this crap has been going on for a month.  
Photo of Stefan Cojocaru

Stefan Cojocaru

  • 8 Posts
  • 11 Reply Likes
It's impossible they are not seeing all this negative feedback.
Switching to aeris tomorrow
Photo of Murphy

Murphy

  • 19 Posts
  • 10 Reply Likes
Does Aeris have any trouble running on Android? We had trouble switching weather providers a while back, but it's getting to be more likely we'll move in that direction. A half day of rewriting code for a new forecast provider sounds a lot better than a bunch of cancelled subscriptions because our product doesn't work right (I really wish WU and TWC felt this way). 
Photo of Paul

Paul

  • 10 Posts
  • 1 Reply Like
Check the date in the response, if it is today cache and display it. If does not match today, display cached forecast. It usually fixes itself after 2-5 hours and the forecast is advertised as updating every 3 hours but this is not always true.
Photo of Stefan Cojocaru

Stefan Cojocaru

  • 8 Posts
  • 10 Reply Likes
We shouldn't have to do this.
This is a paid service after all
Photo of Murphy

Murphy

  • 18 Posts
  • 9 Reply Likes
Agreed. I do want to thank Paul for his workaround.  It's a good idea. It is just crappy that we need to make a workaround. This service isn't cheap in the least. However, the reality of having to rewrite a bunch of code to utilize another service provider has me going in circles. One of our devs just brought up another service that he says is far superior in his experience: Dark Sky. I've heard of it. I don't know much about it. But it looks to be significantly less expensive. I could probably forgive the sporadic errors and need to set up a forecast-versus-present-date check, if it didn't cost around $20 a day.
Photo of Asier Santamaría

Asier Santamaría

  • 15 Posts
  • 0 Reply Likes
same here, today with a lot of requests
Photo of Ulrich Kotitschke

Ulrich Kotitschke

  • 3 Posts
  • 1 Reply Like
Same here. It is always the third of january
Photo of Stefan Cojocaru

Stefan Cojocaru

  • 8 Posts
  • 11 Reply Likes
They just posted an announcement about removing free access to the API.

So they are obviously seeing the problems we are posting, they just choose to ignore them.

Good riddance to this s*itty service.

Aeris is the way to go even if it means spending a few hours adjusting to new API
(Edited)
Photo of Murphy

Murphy

  • 19 Posts
  • 10 Reply Likes
One this page: https://business.weather.com/contact I found a few phone numbers of the parent company. I have actually spoken to a real person about this. I'm sure I'll be exchanging emails with him for the rest of the week before they understand what's up, though.
Photo of Murphy

Murphy

  • 19 Posts
  • 10 Reply Likes
Just heard back from someone at IBM (parent company). Their engineers are looking into it, I've been told.  
Photo of Asier Santamaría

Asier Santamaría

  • 15 Posts
  • 0 Reply Likes
me too
Photo of Jerry Y

Jerry Y

  • 2 Posts
  • 0 Reply Likes

Finally - great news...

Photo of Edward

Edward

  • 7 Posts
  • 0 Reply Likes
This is causing an embarrassing problem for us on many of our WeatherSTEM websites and in our mobile app.
Photo of Edward

Edward

  • 7 Posts
  • 0 Reply Likes
Notice the forecast for this FLORIDA weather station - snow in May!!!
Photo of Murphy

Murphy

  • 19 Posts
  • 10 Reply Likes

Here's one of mine from yesterday. Big temperature swing...Notice also that the forecast starts on Wednesday even though yesterday was Thursday.  
Photo of Ulrich Kotitschke

Ulrich Kotitschke

  • 3 Posts
  • 1 Reply Like
This is because the 3rd of January was a Wednesday
Photo of Murphy

Murphy

  • 15 Posts
  • 9 Reply Likes
I'm gonna quote you back to yourself: It is always the third of January.

I mean, it's kind of humorous. If the end-users could all laugh at how funny it is, I'd feel a little better. Of course, I'd feel much better if WU fixed it, or even responded to us at all.
Photo of Tim Roche

Tim Roche, Official Rep

  • 282 Posts
  • 19 Reply Likes
Hi folks, sorry there are several threads here about this issue. i hadn't gotten to this one yet.  We are looking into the issue.  We have not identified a cause yet.  We've tried a few things to mitigate, and seen some success, with error rates lowering today and yesterday.
Photo of Edward

Edward

  • 7 Posts
  • 0 Reply Likes
It's still happening and, please forgive me because I've been a WU user for way over a decade, but this is seriously ridiculous.  And this has been going on for a very long time - back into 2017.  Here is the output of a WU-provided forecast for a site in northern Virginia for this morning.  I understand the efforts that are underway to migrate WU to other IBM products (many of which I am already using) but some of my legacy code still uses the WU API and I don't have the time or resources to revamp large chunks of my codebase to work around this bug but it looks like that's what I'm going to have to do because the frequency and scope of this issue is increasing not decreasing!!!!!!
Photo of Tim Roche

Tim Roche, Official Rep

  • 282 Posts
  • 19 Reply Likes
Hi all,  I put in another fix for the cache servers today, and from my monitoring, it seems like the errors have stopped and I have not been able to pull an old forecast since the fix went in.

Please let me know if you continue to see errors. 
Photo of dobry

dobry

  • 21 Posts
  • 1 Reply Like
Problem still occurs. We got a claim from customer and I could also reproduce it.
Photo of Edward

Edward

  • 7 Posts
  • 0 Reply Likes
I see it happening now in locations that were fine yesterday.
Photo of Edward

Edward

  • 7 Posts
  • 0 Reply Likes
This is from a site in Florida - unacceptable!!
Photo of Murphy

Murphy

  • 19 Posts
  • 10 Reply Likes
I have at least 2 locations with erroneous data this morning. Renfro Valley, KY, and Mallorytown, Ontario.
Photo of Tim Roche

Tim Roche, Official Rep

  • 282 Posts
  • 19 Reply Likes
Sorry folks, looking some more
Photo of Murphy

Murphy

  • 19 Posts
  • 10 Reply Likes
I've been trying to find a glitched location for the last 20 minutes. Haven't found one. Now looking for (and not finding) the bad information is frustrating me...what a life!
Photo of Paul

Paul

  • 10 Posts
  • 1 Reply Like
Getting better. Yesterday 24 requests 3 errors. So far today 13 requests 1 error
Photo of Ulrich Kotitschke

Ulrich Kotitschke

  • 3 Posts
  • 1 Reply Like
The problem has not been solved. It still exists.
Photo of Edward

Edward

  • 7 Posts
  • 0 Reply Likes
This is the output of a call to forecast10day for Miami, FL - I imagine there has to be hundreds of thousands of people who live in that zone so I imagine this bug is impacting a number of people who use apps that get data from Weather Underground - including ours!
Photo of Stefan Cojocaru

Stefan Cojocaru

  • 8 Posts
  • 10 Reply Likes
It is not related to the location used for the query, I am using European locations and I get the same result
(Edited)
Photo of Edward

Edward

  • 7 Posts
  • 0 Reply Likes
I am getting the result randomly across an increasing number of our query locations - the more populated of an area, the more complaints we field and it seems to be occurring with increasing frequency in an increasing number of locations.