Does Hyperdrive cache in multiple locations, or only near your DB?
Does Hyperdrive cache in multiple locations, or only near your DB?
AMultiple locations. How many is increasing, but we have to briefly pause the rollout of that
Mi'm seeing random hyperdrive error spikes corresponding with 500s for specific routes on my api that i can't reproduce locally
Mit would be really nice to see what these errors are somehow
AWhat's getting fed back to your postgres client? Is there at least useful information there?
We have a track of work for improving error reporting in our dashboard. It's definitely something that needs more polish. Just a matter of balancing priorities.
We have a track of work for improving error reporting in our dashboard. It's definitely something that needs more polish. Just a matter of balancing priorities.
DFor Google Cloud SQL, has anyone here gotten IAM based auth working instead with a Worker?
Dah, seems like I can.. https://cloud.google.com/sql/docs/postgres/iam-logins?authuser=1#logging-in-as-a-user
Dquestion though, how do I tell Hyperdrive that the password is gonna expire after x amount of hours?
Dhttps://developers.cloudflare.com/workers/databases/connecting-to-databases/#_top:~:text=For%20services%20that%20require%20mTLS%20authentication%2C%20use%20mTLS%20certificates%20to%20present%20a%20client%20certificate.
Does this mean I can connect via mTLS?
Does this mean I can connect via mTLS?
Cloudflare Docs
Learn about the different kinds of database integrations Cloudflare supports.
Mhard to tell, i enabled the new cloudflare logging that will hopefully help
Mah yeah there are many 'connection terminated unexpectedly' errors in the worker log
M
DDhmm
DDaren't I already doing it?
AWe currently cannot support SSL handshake from local dev. It's on our roadmap, but a recent attempt to add that support uncovered some pretty serious complications we need to work through.
Dyikes, that's uh, not great. 
So I'm going to have to do my database stuff unencrypted over the internet for a remote database?
So I'm going to have to do my database stuff unencrypted over the internet for a remote database?D@AJR is there a github ticket I can follow for it?
AIt's not my favorite thing right now. You can use
wrangler dev --remotelocalConnectionString but will use the actual Hyperdrive binding you give it and otherwise work like a normal worker.AIt's mostly internal work we think, so no such luck. If you file an issue on GitHub, link it here and I'll see to it that it gets updated when there's updates to be had.
Dhmm, I guess I can try to use cloud-sql-proxy..
DEnded up doing this. Much safer now that I can do local dev with SSL upstream, while wrangler doesn't need to use ssl locally.
IAlso allows IAM auth if you wanted to.
DAren't I still supplying the username/password manually for Hyperdrive?
IYes, but what I mean is that if you use IAM auth locally you can have a crazy long password that's stored only in Hyperdrive (to prevent accidentally leaking it somehow)
DNot sure I understand, I’m already using a crazy long password.
DI would like to prevent leaking it though.
IAny time you have a password used for production infrastructure, using that same password for local development means that it's much easier to make mistakes.
DOh, I’m using a totally different database.
IE.g you forget to
gitignore an .env file, or it appears in your clipboard history, etc.IOh OK, great.
DCould I avoid using a password locally?
Dhmm I think I might be on the right track..
DSeems like CF enforces having a password
Deh, couldn't get
--auto-iam-authn to workDwell, one good news is that with
cloud-sql-proxy, I can remove 0.0.0.0/0 in the allow IPs list in Google Cloud SQL.D@Isaac McFadyen Do you have any opinions on what managed PostgreSQL options are the "best"? Google Cloud SQL seems to be really nice, the cost is reasonable ish, and has near zero downtime even during upgrades.
INot particularely. I use Cloud SQL for another project and it is quite nice, yeah.
DThanks! We use Cloud SQL for two other projects; this is the first time we’re using it for something that runs outside of Google Cloud though. 
GI'm having trouble connecting to my supabase database wether I'm using a pooler connection or a direct connection via hyperdrive. The Hyperdrive binding is
null and the pooler connection times out without ever connecting. Anyone know what might be going wrong?GFixed it. I used an example where the binding was not accessed correctly
AHyperdrive down?
AStarting to see weird behavior for our Workers, which I can't explain myself
N"Connection terminated unexpectedly"?
ANope. Release rolling out right now to address some corner cases with bulk inserts and very large resultsets. But that wouldn't have been going as of 10 hours ago when you messaged.
What kind of weird behavior?
What kind of weird behavior?
DIs there any difference, in terms of programming, between using Hyperdrive vs. TCP Connect with Postgres.js?
DJust wondering if I'm doing any vendor lock-in here.
AIn terms of how you'd actually use them, if you wanted to port away, no not really. You have the option of designing your app more around cached result speed, and also you probably want to avoid transactions more than if you were using direct connections (though that's true with most any connection pooler so.. eh). Overall it should be very portable, as a part of the userbase for Hyperdrive is meant to be folks with pre-existing databases and apps trying to move onto Workers.
If you find cases where that isn't true, please do mention them here.
If you find cases where that isn't true, please do mention them here.
DThanks! This is for new applications, where we want the option to let some enterprise customers self-host (without us needing to rewrite anything).
T@dave | ai.moda the main difference will be using the connectionString you get from Hyperdrive (ie env.HYPERDRIVE.connectionString) instead of env.connectionString accessed as a environment variable/secret.
