Issue downloading sitelink performance report

[Tirtha Ghosh Dastidar]

Hi team,

I am downloading the sitelink performance a.k.a PLACEHOLDER_FEED_ITEM_REPORT.

Half the time it succeeds downloading, half of the time it hangs on one of the last account ids (can send account id in secured email).  After this statement:

INFO:googleads.adwords.report_downloader:Request Summary: {'clientCustomerId': 'xxx', 'skipReportHeader': 'True', 'includeZeroImpressions': 'False', 'server': 'adwords.google.com', 'skipColumnHeader': 'False', 'skipReportSummary': 'True', 'isError': False}

I have tried to identify the problem here, but I just cant see why there would be an intermittent hanging. Nothing is obviously wrong with the code and I would expect the threading approach to work just fine. From my side, however, I cant see any reason why\how this could be a cluster issue.  Since I am running the python app without the mapreduce stack directly from bash and getting intermittent hanging, I am not really sure where to debug this.  

local __DLC_BACKOFF_FACTOR="5"

local __DLC_MAX_RETRIES="5"

local __DLC_PAGE_SIZE="1000"

local __DLC_DOWNLOAD_FORMAT="TSV"

local __DLC_REPORT_TYPE="PLACEHOLDER_FEED_ITEM_REPORT"

local __DLC_DATE_RANGE_TYPE="CUSTOM_DATE"

local __DLC_REPORT_NAME="Custom PLACEHOLDER_FEED_ITEM_REPORT"

local __DLC_FILTER="Status IN ('ENABLED')"

local __DLC_API_VERSION="v201708"

[Dhanya Sundararaju (AdWords API Team)]

Hi Tirtha,

The report download request may hang or time out on extremely large data sets. There is no explicit data size limit, however, the server may return an error if the report is too large. You can use predicates to break up the report request into multiple, smaller requests. For example, use a date range or you could consider submitting multiple requests that each filter for a subset of Campaign IDs. If not already tried, you could use the Parallel Report Download sample code to run the Placeholder Feed Item Reports for all accounts under a MCC in parallel, so that the larger reports do not get in the way for the smaller ones to complete. Also, please note that v201705 and v201708 will be sunset on March 28, 2018. Please migrate to a newer version if you haven't already done so. Please let me know if you have further questions.

Regards,
Dhanya, AdWords API Team

[Tirtha Ghosh Dastidar]

Hi Dhanya,

We are using the parallel download approach. What seems to happen is that for one account (which is one of the biggest accounts), the script hangs and keeps running. I does not fail or error out. We have applied possible predicates but given that we have lots of data, what is a solution here? I have tried playing around with the page size as well. Is there a recommended page size? 

[Dhanya Sundararaju (AdWords API Team)]

Hi Tirtha,

The PAGE_SIZE in the sample code determines the maximum number of account ids returned in the queue for a MCC. So, a change in that value would only restrict a few accounts incase the value set is below the number of accounts in that MCC. If you are looking for pagination, unfortunately, Paging is not supported in reports. However, if you have issue with only 1 account, I would suggest to run Placeholder Feed Item Report for just that account, after applying predicates like CampaignId or week etc so as to break it down further. We will then be able to tackle the issue better. Also, please do run this report via the UI and confirm if it takes as much time to generate. Please let me know if you have further questions.