v13 early sunset ?

[Paul Poulton]

Hi,

Relieved this message this morning.

You are accessing an AdWords API service that has been discontinued.
Calls to this service may fail. Please visit the AdWords API blog for
information on migration to new v2009 AdWords API.

I was under the belief that all versions apart from 201111 were being
sunset on 29th February and today is the 22nd ?

[Paul Poulton]

Apparently:

the core engineering team have started returning a small number of random errors (<10% calls) for v13 reporting to ensure all regular users are aware of the deprecation. This error will be returned for 100% of calls from 29th Feb.

[Ken Foster]

So...if we keep trying the report should eventually work? I've run the report twice today, about 5 hours apart and it failed both times with that error.

[Steve B]

We're getting these too - on what seems significantly more than 10% of calls - more like 25% to 50% of calls in a batch that we do.    It's caused major panic this morning until we figured out what was happening, and is still causing havok in our processing right now.

I'm spending more time right now picking up the pieces from this and reassuring people than I am finishing the migration to the new reports system, which is kind of counter-productive.

I *really* wish they'd A) have given us *some* warning they were going to do this and  B) given us a way (flag in the header or something) to say "Yes, we know about the switchover so don't send us the warnings".   I'd like to see them turn this off until they do both of these things.

It didn't help that I discovered this group is no longer accessible using Safari -- I tried logging in to see what had happened and it kept telling me I wasn't a member, and needed to join the group.   But then when I join, it kept saying I was already a member - and still didn't present the forum.   I had to fall back to using Firefox in a windows virtual machine to see what was going on.

[Bernard Bajoras]

Experiencing the same, including the panic. I've tried fetching reports for one account 5 times and all failed.

[Paul]

Well, I was told...

the average v13 report requires multiple calls to the API, to schedule the job, poll the job, then retrieve the download URL. If c. 10% of calls are failing, and you poll the job 10 times within a single try/catch block then you're almost guaranteeing the catch block will be triggered.

I had trouble accessing the group through Chrome earlier...just got the email notification of the enteries to this thread and could get back in again.

Good Luck all

[Zweitze]

Are you saying they intentionally sabotaged the API servers and therefore our production?

 

I can't believe this.

Anybody from Google care to comment?

[Ed-]

10% error? this is the worst joke ever!

Do you guys know what day is it in brazil today? first working day after carnival.

Do you guys know how many reports they are trying to run this afternoon? Yes, all of them.

AND THE FREAKIN OLD SYSTEM DOES NOT SUPPORT START OVERS!

You effectively sunset the freakin thing 1 week earlier! And I am likely to need to update my cv! :(

[alx]

It's quite frustrating. Our reports fail with at least 50% reate. Google should've had two dates for sunset - one for complete and one for partial.   >:(

[Eric Koleda]

Hi All,

I apologize for not jumping on this thread sooner.  Paul is correct, we have enabled these deprecation errors on about 10% of all v13 requests, as a way to notify developers that have failed to migrate.  As he mentioned, it may take many requests to download a report (with polling ,etc) so you should you implement retry logic on a per-request basis, rather than a per-report basis.  We apologize for any panic this may have caused, and we're here to help you with any migration questions you have.

Best,

- Eric Koleda, AdWords API Team

[Jim McCabe]

Eric -

Will this also affect the v13 reseller API that is remaining after sunset?  I'm referring to AccountInterface.addBudget() - we are looking for assurance that:

a) This API will remain after the sunset.

b) This API will not experience these intentional failures.

Jim

[Tilendor]

We have also been blindsided by this.  We have been working on the new API code right now and are needing to release early.  Bad form Google.  This would have been acceptable if there had been notice. 

Blindsiding people who are already aware of and working against the published deadline is a bad experience.  I echo the comment of having a 'partial cutoff' and a hard cutoff.  Maybe don't do anything with the API until the day after sunset, and then have a grace period with this 10% API errors idea.  Starting this early and unannounced is unprofessional for a business that knows other business rely on the API.

I wonder how many companies that are already working on the new API have been sidetracked by this.

[Ken Foster]

FWIW, I had a 2 minute sleep between polling tries on a report download and it failed with the error every time eventually. I changed my polling sleep interval to 5 minutes and got it to work finally. Reduce the number of Pollings and the chances of being in the 90% get better.

[j.e.frank]

I'd like to add my voice to the "bad decision" chorus.  The intent seems good (making sure everyone knows about the sunset) but it's such an unexpected and heavy-handed way to do it.  We've been getting hammered with messaging about the Feb 29 sunset, which is great.  My interpretation was always that the previous API versions *would not* be available after Feb 29, but also that they *would* be available prior to Feb 29.  This is pulling the rug out.  I don't expect anyone to try to defend the decision, and like I said I understand and appreciate the intent, but it was just not the right way to do it.  I just hope Google will learn from this and not do something like that again.  It's hard enough for us developers to keep up with the API changes, don't make it harder.

[Paul Poulton]

we are in a similar position to many of the responders to this thread.

The data retrieved via the API is a dependancy in our data warehouse, so when it fails someone has to get up at 4am and look why, and do something so our build continues.  Essentially it is only a 3rd party data feed, but one that has caused 40% of our failures in the past 3 months.

We didn't manage to retrieve any data yesterday which means I miss a day of parallel testing.

Today we have done 10 attempts and still not retrieved any data.  I guess I will download it via the webbie.  Time which I had planned for remaining Development and Test.

Just an Idea, Bing and Yahoo are merging their API, their approach was to send an email asking for a response if you are ready for the migration.

Any chance you could turn this 'feature' off until 29th now because I have plenty of error messages to go by now, you can consider me warned thank you very much.

[Rori Stumpf]

I'd like the fake error messages to stop.

It's not necessary to pop these surprises in the middle of the night.