[∞] Performance issues

[nmorduch]

• Slow loading times
  • weekly page took 36 seconds for everything below the main toolbar to load whereas the main Publications page only took 5 seconds
  • The content grid on the Event page took 17 seconds
• Memory overload for web dynos
  • I just decreased to 5 workers but not sure this will help
• System exits in Sentry might be related?

[kaedroho]

Naomi and I had a look through New Relic during our catchup today:

• The Weekly takes an average of 7 seconds to load. A lot of time is spent in Python and template rendering. It runs about 150 queries for images so I think it's most likely to be image processing causing the issue. We don't think this view should be fetching any images though because the listing is rendered with React so this might be an easy fix
• The Meta API is called on every single request and takes over a second each time to respond. If this is returning the same response for every user we can cache this.
• The system exits are because the Gunicorn workers are timing out. I think there's a chance this might be reduced/eliminated when we fix the meta API (which currently uses up around 50% of total request time on its own).

[kaedroho]

We had another look at this. There's also a performance issue with editing reports. Sometimes the page editor times out and the page can't be edited.

According to New Relic, It's spending a lot of time rendering streamfield blocks. Switching to wagtail-react-streamfield might solve this.

[nmorduch]

Search still has problems, probably fixed in #1494. Post api is still having problems.

[nmorduch]

Changed to medium priority because we've made some progress on this but there's still more we could do. ie it's still bad but not abominable.

[nmorduch]

This is not worth it in the near future if we're redoing our FE