Today released break prod

Today i wanted to try the new logs system and added it in the wrangler config.

Then deployed and the workers cannot connect to my db anymore.

So i removed the setting and publish again.

But the problem persist.

My only solution was to revert the release from the dashboard.

texan•9/26/24, 9:38 PM

Hey Martin, we had an outage with workers that may have affected your deployment:
https://www.cloudflarestatus.com/incidents/1lrs7s9vc8lv

Workers issues on workers.dev domains

MartinOP•9/26/24, 10:05 PM

Thanks a lot i will check but i’m not sure itms the same issue, i have custom domain, or maybe i don’t understand the scope of the issue ^^

MartinOP•9/26/24, 10:14 PM

I still get the same error:
PostgresError: Tenant or user not found
When it try to connect to supabase DB
The previous deployment do not have the issue

MartinOP•9/26/24, 10:23 PM

Here is endpoint who just try to connect to the postgres db and fail in the preprod ( with the exact same git checksum deployed )
https://api-alpha.capgo.app/private/latency_drizzle
In prod with reverted deploy it work:
https://api.capgo.app/private/latency_drizzle
The code of the deploy:
https://github.com/Cap-go/capgo/tree/9aabc6f518a687e1e11174b875025d2671d1ef79

GitHub

GitHub - Cap-go/capgo at 9aabc6f518a687e1e11174b875025d2671d1ef79

Instant update for capacitor apps. Contribute to Cap-go/capgo development by creating an account on GitHub.

MartinOP•9/26/24, 10:23 PM

@Will Taylor | Workers i hope that can help

texan•9/26/24, 10:27 PM

Ah yeah, definitely a different issue. I’m not very well versed with that sort of setup, might have to wait for someone else to reply

Walshy•9/26/24, 10:45 PM

That is definitely an issue within your config/code - I'd recommend verifying the tenant/user does exist in postgres/where this is hosted and figure out if the dns and things are all correct. Make sure no tokens/auth have expired

MartinOP•9/26/24, 10:55 PM

@Walshy | Deploying the exact same checksum code is deployed on both, one work not the other one ( who is a revert from yesterday)
All the config is 100% versioned.
I wish it will my code i looked for 4 hours until i just tested to deploy the same checksum, and found out this.

Walshy•9/26/24, 10:58 PM

Same code does not mean something did not change at postgres

I'm not sure how anything our side could possibly make a postgres user not exist haha

MartinOP•9/26/24, 11:04 PM

I agree that odd, but the same env var is used, and wasn't change between yesterday and today.
And supabase didn't said they deploy anything lately.
Especialy on the DB side i control the versionning

MartinOP•9/26/24, 11:24 PM

Ok i found the difference, It was in the ENV var you where right @Walshy | Deploying sorry for the disturbance

Walshy•9/26/24, 11:24 PM

Haha you're all good - glad it's resolved!

texan•9/26/24, 11:48 PM

Walshy is the GOAT!

MartinOP•9/27/24, 1:44 AM

Clearly

Today released break prod

Similar Threads

Similar Threads

Similar Threads