Assuming this is a Google account with enabled billing, how do you prevent abuse? Very conservative rate limits? Prepaid credit card (is that even an option)?

2 secret limit per subdomain. The thing that runs the serverside cells is rate limited to 1 VM max. If there is genuine interest I will add the ability to top up with premium credits.

Looks like this page is having some errors -- ui = RuntimeError: View is not defined

whoopsies, over eager refactor earlier removed a critical dependancy. Fixed now.

Seems this notebook has again become the victim of some refactoring :)

Works for me but it's slow as hell at resolving

Ah, right. The "next." subdomain hasn't been whitelisted, causing the error "ui = RuntimeError: Not allowed to be embedded".

fixed, it was firebase dependancy that broke with next. (broke from the anti-clickjacking fix, *sigh*)

Just a note, it seemed like the promise for the `user` variable wasn't getting resolved for some reason, so I just set `user` to `viewof modernUser.value` and that fixed the issue.

thanks for the report. I think I fixed it.

So this isn't working for me at the moment, but I think it's a larger issue with the deploy service today. Basically it looks like I'm getting a 504 timeout ('upstream request timeout') when I try to access any deployed function (including the fetch calls in this notebook and the trivial deploy in the quickstart notebook). I don't see any issue on the latency monitor, so maybe this is a problem with my IP? I was able to set a secret successfully the other day. Would be happy to help with debugging via Zulip or wherever is convenient. It's probably just a symptom, but I also noticed CORS complaints for this notebook in the console : "Access to fetch at 'https://webcode.run/observablehq.com/@endpointservices/secrets/' from origin 'https://endpointservices.static.observableusercontent.com' has been blocked by CORS...").

Ahem, yeah the whole serverless cells were down. I changed permissions on something for a new feature and it invalidated some old assumptions that were still in prod. The prober captures the 5 hour outage nicely: https://observablehq.com/@tomlarkworthy/serverless-cell-latency-monitor I will try to do better. I need an alerting system for one thing...

This works now! Thanks for the quick fixes, this system is super super cool. Huh, I totally didn't see that gap until you pointed it out... I guess I just didn't realize the times on the graph were not in my local time and the outage happened to start at a local time similar to the local time when I checked.

yeah if there are no points on the graph it does not include "now" so it just looks like its working. It's not the greatest dashboard I also just fixed the clear performance regression from the build a couple of days ago. That said, the latency prober gives constant traffic so I can definitely alert on "lack of 200 responses" in a 45 min window. That would of caught today effectively. I totally was not expecting an outage as I never actually deployed anything. FYI: The feature I am working on is gonna be cool x 100 though. I am so excited. The 5 hours of downtime will be worth it!

auth code: DDssjVYZpxq2pE_xrFiP

auth code: Szp0kWgWVQXb5qOwzJpT

auth code: Ayd6oMEqyhYZPPJJMOlE

rX8ACoDMubpR_13uz38Y

auth code: rX8ACoDMubpR_13uz38Y

public auth code: rX8ACoDMubpR_13uz38Y

public auth code: F4_UVZXAGBM4IpZGOs3G

public auth code: G8qSiIo5Oe6oHjQL97ZD

public auth code: q0TA7ilrtcf8umZu_Or_

public auth code: Bh4TOw7lfE9Fyt6KBACB

public auth code: odnyzsahnyHPZVXsePFP

public auth code: QiOMzAYkjxlzL2Y0s8qj

public auth code: WbBSSTdFCvbpb_1t_LNV

public auth code: V2KyVboNEMtfWuI0Q2pr

public auth code: SmI8JWcXxqcH0PlHvty3

public auth code: ZJ0eTIqqFAHrqNB5rUK2

public auth code: y23bjFKgQEIbt6oML9KU

public auth code: fC5EUVN__7qNf3Znevga

public auth code: PpKYtU6t0Vrp1UyJ6xu5