oh, boo, there goes my entire strategy

keep in mind it will log every request from the workers, not just the ones that would have been blocked

I have some local code i thought i'd move to workers now that they support postgresql.. but.. tunnels dont support postgresql 😛

if I disable it, will it log the blocked ones? (I assumed not.)

All blocked requests are logged

fwiw it might make more sense to use something like an http api (I believe postgres has an extension for it directly) anyway, as spinning up new connections is expensive, and you'd want some sort of connection pooling. Implementing that yourself with DOs is kind of messy

solutions like PostgREST make this a piece of cake

If it got blocked, it didn't get skipped by the rule

every request from the Workers that target my own zone, right?

yea already got the API, the idea was to move the http api to cloudflare. thought it'd be a nice step into migrating to workers

yes indeed, anything that matches the rule criteria

ah cool, I don't think I do that many self-referencing fetches

feel free to log then, might be fun to see

enabled

my security events is a mess without filtering out all the skip logs

I already have over 1,400,000 logpush objects

31.3k firewall logpush objects

Workers are by far the worst "offender" (but I like it 😄 ) for logpush.

yea i was thinking a DO.. basically following this high level.. but i guess they assume the database is exposed to the internet. https://developers.cloudflare.com/workers/databases/connect-to-postgres/

Logpush is awesome, so useful

my workers logs go to Sumo Logic which is fun because i can then search them for errors and it will actually tell me why the errors happened (within the best of its abilities)

looks like it. In that blog itself they recommend using something like Pgbouncer. There was an old abandoned modified postgres driver to use tcp websockets/cf access, but I wouldn't recommend it

yeah 0 hits so far

@Chaika yea i looked at the code of that. it was kinda odd 😄

now I'm confused why I thought this was a problem I needed to fix?

something fun I've noticed with logpush is that even with frequency low, workers trace events are batched/pushed in way smaller amounts

yeah the HTTP logpush files are much bigger, with fewer objects

hmm, this doesnt seem like it would work either.. unless the worker had a tunnel.. https://blog.cloudflare.com/cloudflare-tunnel-for-postgres/

DNS gets more raw data I assume, but also just seems to batch more / more infrequent