| Krinkle | |
| Jul 12 2019, 6:29 PM |
| F29750759: Screenshot 2019-07-12 at 19.24.40.png | |
| Jul 12 2019, 6:29 PM |
| F29750795: Screenshot 2019-07-12 at 19.26.10.png | |
| Jul 12 2019, 6:29 PM |
| F29750754: Screenshot 2019-07-12 at 19.22.30.png | |
| Jul 12 2019, 6:29 PM |
| F29750750: Screenshot 2019-07-12 at 19.22.49.png | |
| Jul 12 2019, 6:29 PM |
Here seems to be a fairly confident +50ms regression on page load time (at the 75th percentile).
Found through navtiming-loadEventEnd, but also seen under:
See also https://wikitech.wikimedia.org/wiki/Performance/Metrics#navtiming-2_metrics for the definitions of these metrics.
| Dashboard: Navigation Timing by Platform (desktop) |
|---|
 |
Note: The CSS that spiked in T227929 was subject to caching however SWATs on 15th and 16th would have reduced the CSS payload for most users.
https://grafana.wikimedia.org/d/000000326/navigation-timing-alerts?refresh=5m&orgId=1&panelId=19&fullscreen&from=now-7d&to=now does seem to be in decline now @Krinkle - do you think this was the cause?
