I’ve had this persistent on and off again problem where my backup server in my homelab will just randomly stop responding after a few days of operation.
The specs are:
- Intel Core i3-8100
- 32GB of DDR-4 RAM (originally 16GB)
- 5x 6TB Seagate Ironwolf in a Storage Spaces two-way mirror (previously a hardware RAID6) using ReFS
- Windows Server 2025 (this issue happened in 2019 and 2022)
The behaviour was weird. Everything would work fine for a few days and then the server would be inaccessible via RDP or the local console. It still ping’d and RDP would get to ‘Configuring network connection’ but never complete. Veeam backups would also stop.
I had read that when using ReFS in Windows you want roughly 0.5GB of RAM for every 1TB of storage. My server originally had 16GB of RAM so I figured this must be the problem and upgraded to 32GB which did not resolve the issue.
I figured the issue had to be disk related, don’t ask me how I got this hunch, and then I found this old thread: https://forums.veeam.com/veeam-backup-replication-f2/refs-to-defag-or-not-to-defrag-that-s-the-question-t43401.html
Turns out, by default, Windows Server automatically defrags or “optimizes” disks on a weekly schedule.
To disable disk optimization follow these steps:
- Click ‘Computer’
- Right click the ReFS volume and choose ‘Properties’
- Go to the ‘Tools’ tab
- Click ‘Optimize’
- Select the ReFS volume and click ‘Change Settings’
- Uncheck ‘Run on a schedule’ and click ‘Ok’
- Click ‘Close’, ‘Apply’ and ‘Ok’
Since disabling this on my backup server I haven’t had a single crash/hang.
Based on that forum post, it sounds like Veeam does not recommend running this optimization process on ReFS volumes used as Veeam backup repositories anyway. I wish their installer checked for this setting and warned you. Then again, how many people run Veeam on a Core i3 with consumer hardware?