adunkman
commented
6 years ago Looking at the data, requests that are triggering timeouts are almost always requests for individual service requests. There are 604 requests for service request numbers in the last 24 hours:
| Response Time | Number of Requests | Percent of Requests |
|---|---|---|
| 0-1s | 576 | 95.4% |
| 1-5s | 6 | 1% |
| 5-29s | 10 | 1.7% |
| 29-31s* | 12 | 2% |
* Heroku’s response timeout is 30 seconds (sometimes 30.001 seconds).
As discovered through logs, DC’s 311 endpoint sometimes doesn’t return data for… 30+ seconds. This isn’t an ideal experience for our service (it looks like this app isn’t working well).
I’m not sure what the best thing to do would be — I can think of two solutions:
Timeout after a certain amount of time, and show an error that says DC’s 311 service is too slow.
Timeout after a certain amount of time, and show an error that says we’re still requesting data from DC’s 311 service, and then client-fetch the data.