Strange CheckDB Results

Question

I think I have a failing disk sub-system on a Proxmox VM running Microsoft SQL Server 2019 (RTM-CU32-GDR) (KB5065222) - 15.0.4445.1 (X64) Standard Edition (64-bit) on Windows Server 2022 Standard 10.0 (Build 20348: ). I have no visability on how the VM is configured.

The Windows log shows Warning Event 129, viostor. Reset to device, \Device\RaidPort2, was issued.

Within SQL Server read and write waits are about the same as the baseline except for when an occasional 825 error occurs.

What I am finding confusing is the CheckDB results. Running:

DBCC CHECKDB (MainDb) WITH NO_INFOMSGS, ALL_ERRORMSGS, DATA_PURITY;

This produces, 26 consistency errors in one table at 21:00 and 15 consistency errors in another table at 06:00 with no sign of any errors in the original table. There are never any allocation errors. I am unable to publish the actual CheckDB results.

I would be grateful for any ideas on what is happening.

Answer 1

I think you got this. I see no other way interpreting these symptoms as something under SQL Server/Windows failing. But I don't think you can do anything on your own. You need to raise this issue with people operating the hardware, Proxmox and make management aware. Right now it looks like SQL Server may be delivering corrupt data and even restoring database will not help you because the underlying storage cannot be relied on. But you should start testing your backups, if they are usable or got corrupted too.