I recently upgraded to a 28TB parity drive, and I’m trying to dial in the best practice parity check schedule.

Current behavior:

• Full parity check takes ~2.5–3 days to complete

• System remains usable, but I obviously want to minimize wear and performance impact

Questions:

1. With modern large drives (28TB+), what is the recommended parity check frequency?

   • Monthly?

   • Quarterly?

   • Bi-annually?

2. Do most of you still run non-correcting parity checks routinely and only run correcting checks when errors are detected?

3. Any downside to spacing checks further apart given how long they now take?

4. Are there any Unraid-specific tunables settings or other plugins you recommend to optimize long parity checks on large arrays? Currently, using Parity Check tuning plugin.

I’m mainly looking for a balance between data integrity and drive longevity, especially with parity checks taking multiple days now.

Appreciate any insight from others running 20TB+ / 28TB parity setups.

1-Quarterly

2-Yes

3-No downside in a small home use scenario like mine

4-No

3 hours ago, Veah said:

1-Quarterly

2-Yes

3-No downside in a small home use scenario like mine

4-No

Agree with all of the above. That is exactly how my main server is configured (no parity in the second server).

I support this if the server is stable and has a history of being issue free. Mine runs with UPS power, has all "can't lose" data backed up regularly, and historically has uptime measured in several months at a time. I did run a manual parity check after updating to Unraid 7.2.0 for peace of mind and not seeing that I had cancelled the last run check.

Note I don't use the Parity Check Tuning plugin. My check runs ~26 hours for 18TB. I schedule it for a time where user and maintenance tasks are typically low, and really haven't seen any impact using the server when the check is running. There are 20 Docker containers and 2 small VMs running on it 24/7/365.

The longer the interval between parity checks, the greater the need for the system administrator to actually spend time monitoring what is going on. With a longer interval and a neglectful admin, the higher the probability is that the array will eventually have more failures than the parity can recover from. (If the system is not monitored actively, a disk failure can go undetected for months as parity can cover for it With the speed of modern disks and hardware, parity-build data recovery is virtually transparent to the average user!)