Closed Skillz4Killz closed 7 years ago
This is what we get on an AWS server in us-east-1, averaged per 3h periods. There is a gap from when our monitoring was offline.
Edit: I was able to recover the logs from a backup, here is the graph with the gap closed (but without the most recent data):
The raw spreadsheet (first graph's data): https://gist.github.com/schneefux/67b2bd6fcccf8fac5eba8205306b64c1 The average minimum response time is 80ms (404 no matches). The average average response time was 250ms until a few weeks ago, now it is 600ms. The average maximum response time was 30s until 07-25, now it is 60s (504 gateway timeout).
We deployed a few big backend changes at the end of July, but it looks more likely to me that the API became slower after the "small upgrade" on July 25th (also in regard to #248).
Thank you Shutter appreciate it
hey dudes - just jumping in.. if you're seeing 60s then something is clearly wrong. It should be <1s in my mind. The caveats are - some data is large, so we should take transfer time into account and be focused on time to first byte.
@genexp Do you have a recommendation for a client side request timeout?
I just set up logs for the time to first byte, thank you for the pointer.
What is considered average response times for the API?
1) Good Response Time 2) Slow Response Time (possibly due to high amount of matches) 3) Super Slow (Something is wrong here)