Announcement: Performance issues with analytics endpoints since 2017-08-15

announcements
analytics
async

#1

We are actively investigating the root cause and mitigation of degraded performance of both the async analytics endpoint POST stats/jobs/accounts/:account_id and synchronous analytics endpoint GET stats/accounts/:account_id.

The latency and amount of 503 errors significantly increased at 2017-08-22 05:00 UTC. Urgent action is being taken to decrease the latency but no ETA is available for complete resolution.

The asynchronous endpoint has had spikes of degradation since 2017-08-15 02:00:00 UTC. Due to the nature of in-order processing, a backlog may be created for queued jobs. In case of any issues with async jobs stuck in PROCESSING state, the recommended path is to use the DELETE stats/jobs/accounts/:account_id/:job_id endpoint and retry once the issue is resolved.

This thread will be updated with the latest status until the issue is completely resolved.


Analytics API service availability (HTTP 503 and very long-running async stats reports)
Pending Twitter jobs for hours
Twitter Ad Stats Jobs Timeout
Analytics jobs finishing slowly or not at all
Ads API is incredibly slow and/or times out
Analytics asynchronous endpoint cached?
#2

#3

We have deployed a few changes that have increased success rate and decreased latency. These measures have been relatively stable since about 2017-08-22 16:30:00 UTC. We are still actively monitoring these endpoints and will continue working toward improving these performance issues.


#4

Both success rate and latency returned to expected levels on 2017-08-24 22:00:00 UTC and have been stable since.

We appreciate your patience as we worked to resolve this.