I would like to provide an update for ticket SU-485200. The database is still unavailable. The project dashboard reports ACTIVE_HEALTHY, but PostgreSQL continues to return: 57P03: the database system is not accepting connections

[Kissa19]

Hot standby mode is disabled.

I also received an automated warning that this project is depleting its Disk IO Budget. We have stopped the backfill workload and are avoiding additional database traffic, but the database has still not recovered.

Could you please check the database instance/recovery state?
Thank you.

[k1ng-arthur]

Have you tried turning off the data api, waiting for a short while and then restarting the project?

[Kissa19]

Thanks for the suggestion. I have already tried pausing/restoring the project once, but I have not tried disabling the Data API before restarting it. The database contains important data and currently returns 57P03: the database system is not accepting connections / Hot standby mode is disabled. Before I try this, could you please confirm that disabling the Data API and restarting the project will not reset, recreate, or delete the existing PostgreSQL database/data? ในวันที่ เสาร์ 26 ก.ย. 2026 เวลา 12:39 Ibrahim ***@***.***> เขียนว่า:

…

Have you tried turning off the data api, waiting for a short while and then restarting the project? — Reply to this email directly, view it on GitHub <#50933?email_source=notifications&email_token=BVZ2OIW6BKG4FTEKX2W7JVL5Q5I7XA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBWGA4DSOJVUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVRTG633UMVZF6Y3MNFRWW#discussioncomment-18608995>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BVZ2OIU37XDJCMQN2N73ZKD5Q5I7XAVCNFSNUABIKJSXA33TNF2G64TZHMZDCNBVHA3TCOJTHNCGS43DOVZXG2LPNY5TCMBYHEYTKNRVUF3AE> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/BVZ2OIWNMERYJ2YVSKHQPU35Q5I7XA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBWGA4DSOJVUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVJTG633UMVZF62LPOM> and Android <https://github.com/notifications/mobile/android/BVZ2OIX7VMAVMVFM2K4WQJD5Q5I7XA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCOBWGA4DSOJVUZZGKYLTN5XKMYLVORUG64VFMV3GK3TUVZTG633UMVZF6YLOMRZG62LE>. Download it today! You are receiving this because you authored the thread.Message ID: ***@***.***>

[BryceEWatson]

To your question first: no, turning off the Data API and restarting won't delete or reset your database. The Data API is only the REST layer (PostgREST) in front of Postgres, and a restart stops and starts the same database on the same disk. Your tables stay where they are.

That said, I'd hold off on more restarts or pause/restore cycles until support replies. "Not accepting connections" with "Hot standby mode is disabled" is the message Postgres gives while it's in recovery, replaying its log before it lets anyone connect. With the Disk IO budget used up, the instance is throttled, so that replay can be slow, and a restart interrupts it.

While you wait, three checks that are safe to do:

1. Observability > Database Health in the dashboard shows your Disk IO budget. Also check how full the disk is. A backfill can fill the disk, and a nearly full disk can cause this exact error (discussion 45958 shows the same message at 92% disk). If it's close to the limit, put the number in your ticket, because then the fix is more disk, not a restart.
2. Database > Backups, if you're on a paid plan. See which daily backups exist so you know your safety net. Don't start a restore into this project while it's recovering; ask support first.
3. Your ticket, SU-485200. Add the time the backfill stopped, the time of your pause/restore, and the Disk IO warning, so they can find the right window in their logs.

A bigger compute add-on gives you more disk throughput, and Supabase's Disk IO guide suggests it, but the change comes with a few minutes of downtime, so I'd only do it once support confirms the database has finished recovering or tells you it's the fix: https://supabase.com/docs/guides/troubleshooting/exhaust-disk-io