AdHoc Reporting Service Errors

[pete]

As of 19:00 GMT this evening, we've been getting errors when requesting AdHoc Reports, (Ad and Keyword Performance report types). 

In v201206 the errors is 

    "AdHoc Report Definition Request failed: !!!1|||-1|||Could not read report???"

In v201209

    

The error is occurring across all the Accounts we manage.

We've not made any code changes and other API requests within the jobs are working fine.

Is there a problem with the AdHoc Reports service? 

[oseemann]

We see a similar problem. For about 10% of our managed accounts we receive ERROR_GETTING_RESPONSE_FROM_BACKEND errors, apparently consistently.

Oliver

[Abhinay Mehta]

We are also seeing HTTP 500 errors this morning when trying to generate Keyword Performance reports for some of our accounts.

[oseemann]

It appears as if the timeout between api frontend and backend has been drastically reduced. Whereas before (on some cases with very large report sizes) it used to take up to 5 minutes for the API frontend to wait and then return the error ERROR_GETTING_RESPONSE_FROM_BACKEND, it now only takes approx. 20 seconds.

Hence we see a lot more of those backend timeout errors. And where before a retry after a few minutes generally succeeded, all subsequent retries now fail as well.

The problem with such a low timeout is, we cannot even download a list of all adgroups for some accounts, esp. where adgroup count is >100k.

Can someone from Google confirm this new timeout policy? Or is it just a temporary glitch that will be fixed?

[pete]

We're still getting these errors.

I've migrated to use v201209 from v201206 - other than a different error message, still can't reliably use the AdHoc report service.

I.e.

<reportDownloadError>

<ApiError>

<type>ReportDownloadError.ERROR_GETTING_RESPONSE_FROM_BACKEND</type>

<trigger>Unable to read report data</trigger>

<fieldPath/>

</ApiError>

</reportDownloadError>

The only success I've had is to limit separate AdHoc requests to a single Campaign (when requesting AdGroups, Ads and Keywords reports), wrapped in a retry if the error occurs.

Just love Friday code-pushes!

Can anyone from Google please let us know what the issue is here? We've been getting the errors intermittently for 20 hours now. 

pete

[Abhinay Mehta]

Could someone from Google please tell us what is going on? Most of our reports for the 17th of Jan are still failing, we need this data.

On Thursday, January 17, 2013 11:05:07 PM UTC, pete wrote:

[Abhinay Mehta]

Sorry I mean most of our reports SINCE the 17th of Jan are failing.

[Michael Ni]

same thing over here, what is going on at google?

[chris h]

Had same problem here.  My solution was to switch to OAuth 2.0 - that is working for me.  So it must be something Google changed with the authentication.

[Takeshi Hagikura (AdWords API Team)]

Hi, 

The temporal issue was happening in our side.

The issue has been fixed now.

If you are still seeing the error, could you please post it again?

Best,

- Takeshi, AdWords API Team

[pete]

The last occurrence in our system was at 21st Jan 2am UTC.

That said, I found a workaround by limiting the requests to no more than a single campaigns objects, so I wouldn't expect to see too many errors of this type anyway.

How come it took you guys the best part of 4 days to even acknowledge this thread?

pete

[oseemann]

Was this change an unintentional mistake that is now fixed or was it an intentional change that, sort of, had unforeseen consequences and has been rolled back?

I'm asking because we have developed workarounds over the weekend (splitting report downloads by campaign ids) and the answer might be helpful while deciding whether to keep the workarounds around or not.

[Takeshi Hagikura (AdWords API Team)]

Hi 

The latter is the case.

That was intentional change, but has been rolled back.

But in either case, it should be fixed now.

Best,

- Takeshi, AdWords API Team