-
Oh I didn't realise you were already using it - I could have just looked at the codebase :/
Congrats on the upgrade - did you see any improvement in response times and cpu usage, etc?
cyclotron3k-
Fractional benefits. This site was been optimised to hell about 8 years ago and hasn't needed to be revisited, and so the benefits from upgrades isn't significant.
That said, I am going to book a week off work to spend time working on the site and I want to add a lot more observability so that I can get a far better idea of slow queries and where the latency is. At the moment my hunch is that nearly all of the time is spent in the Django / Python rather than in the database or Go (API) layer. That hunch is driven by my local dev environment, where I have partially implemented the web layer in Go and the speedup is phenomenal and it's really revealing how bad the Python layer is in comparison.
The SQL layer... I have a slow query log and virtually nothing goes into it except for some very generic searches on the largest sites.
Velocio
We are in the latest postgres 14.
And yeah the search is phenomenal.
The advantage of implementing search in the database are twofold:
We're using the search in a pretty advanced way, every single feature is being used from weighting, ranking, result snippets. It's pretty cool.