Now that all my HDDs (4x 20TB WD Red Pro) have arrived and I’ve run and initial burn-in test (full write and read cycle), I have created my first pool (“elephant”) in my homelab system running TrueNAS Scale 24.10.1 with the four drives in raidz2 configuration. (For hardware info see my signature.)
After configuring mail and alert settings, I am now looking at setting up data scrubs and long SMART tests and I am wondering what a suitable frequency would be. Originally I was planning to run each job on a monthly basis (2 weeks apart).
I’ve never used such large drives and - despite the relatively large capacity - they are also rated at an annual workload of 550TB like several of their smaller capacity models (e.g. the 2TB model).
So if the pool is filled up 50%, a monthly scrub (presumably only over used blocks) would be 120TB/year and a monthly long SMART test (presumably across all blocks) would be 240TB/year for a total of 360TB/year. (I did ignore TiB vs TB for this back-of-the-envelope calculation.)
In sum those data maintenance jobs would take up a little less than 2/3rd of the annual workload, which seems quite high. So I am wondering, if it would make sense to reduce the frequency of the long SMART test to every other month, instead of monthly (and keep the scrub at a monthly frequency). Does this make sense, or am I potentially inviting disaster that way?
A side note re. WD Gold series
I see that the WD Gold series doesn’t have an annual workload rating listed in the datasheet, but even in the Gold line some of the reliability calculation is based upon this 550TB/year workload (footnote 4). So it would seem that even in an enterprise setting this might be a potential issue.
How do others deal with this conundrum and how should I best deal with it?