TrueNAS on Old Hardware: Recycle or Buy New?

When a server rotates out of production after five or six years, it rarely goes straight to the recycling bin. The typical question from mid-sized businesses is: “Can we not build a TrueNAS out of it?” The short answer is a clear “it depends.” There are good reasons for recycling — as a backup target, as secondary storage, as a test system — and equally good reasons to buy new for production. This article brings both perspectives together and offers a checklist you can use in your own server room to make a defensible decision.

In our TrueNAS projects with SMB customers we regularly see both extremes: the pure “keep using it because it is still there” without any look at operating costs, and the reflexive “buy everything new because the vendor says so” without regard for ecological or financial sense. The interesting territory lies between these poles.

ECC RAM is not negotiable

Before we talk about age, efficiency and warranty, there is one hard requirement: ECC RAM. ZFS lives from its integrity checking — block-level checksums, self-healing through redundancy, copy-on-write. All of that assumes that the data in main memory is correct while ZFS calculates and writes it. A bit flip in non-ECC RAM means ZFS writes incorrect data with a valid checksum onto the disks — and cheerfully confirms during the next scrub that everything is fine. The error becomes silent, but persistent.

In practice: The old workstation from marketing, the decommissioned consumer motherboard, the mini-PC without ECC support — all unsuitable for a serious TrueNAS. An old Xeon-E5 or Xeon-D rack server, on the other hand, brings ECC support from the factory. That is the first filter, and it is binary.

When recycling makes sense

A server rolled out of production between 2018 and 2021 is almost always still fit for workloads where low latency or maximum IOPS do not matter. Concretely:

• Backup target for Proxmox Backup Server or Veeam. The load consists of large sequential writes and the occasional restore. An old 12-bay chassis with HBA and 8-to-16 TB SATA disks delivers meaningful performance here at a fraction of the new price.
• Secondary storage and archive. Cold data, compliance archives, ISO repos, software depots. A RAIDZ2 pool of older disks is sufficient, as long as regular scrubs run.
• Test and staging environment. For restore tests, ZFS replication targets or trying out new TrueNAS SCALE features, recycled hardware is ideal — an outage does not disturb operations.
• Off-site replica. A second location needs capacity, not performance. An old server receiving a few terabytes per night via zfs send does exactly that.

The common denominator: The data is secondary or tertiary, recovery objectives (RTO/RPO) are relaxed, and a system outage at most extends a backup window, it does not halt operations.

When you should buy new

In production, for the primary VM datastore role, or as the NAS on which 50 users have their file shares, the math is different. Four points matter:

1. Warranty and spare parts. A server without active manufacturer warranty is a problem when it fails. Third-party maintenance providers exist for used enterprise hardware, but SLAs are not always identical to what the manufacturer delivers.
2. PSU efficiency. A 750 W 80-PLUS Silver PSU from 2016 in reality often runs at 82 to 85 percent efficiency in the typical load range. A current Titanium PSU is at 94 percent. In 24/7 operation the difference adds up over a few years to amounts that justify hardware purchases.
3. Platform support. Older HBAs (LSI 2008) still run with current TrueNAS versions, but Broadcom discontinued support long ago. For production systems, a supported platform (LSI 3008 or newer, Tri-Mode HBAs for NVMe) is the safer choice.
4. Energy and cooling costs. A modern Xeon-D or EPYC Embedded server with the same capacity often draws half the wall power compared to a dual-socket Xeon-E5 from 2017.

Rule of thumb from our TrueNAS projects: As soon as the server is visible to users or VMs in terms of latency, it belongs in the “new” category.

The checklist in detail

When a candidate is in front of you, we work through these points:

Criterion	Green	Yellow	Red
Year	2020 or newer	2017—2019	older than 2017
ECC RAM	present, registered	present, unbuffered	not present
HBA	LSI 3008 / 3408 in IT mode	LSI 2308 in IT mode	RAID controller without IT mode
Backplane	SAS3 12 Gb	SAS2 6 Gb	SATA direct-attach
PSU	80-PLUS Platinum / Titanium, redundant	Gold, redundant	Bronze or single PSU
Disks	new / SSD / lightly used	used enterprise below 30k hours	SMR consumer or above 50k hours
Warranty	active manufacturer contract	third-party available	none
Boot media	SATA-DOM or NVMe (mirrored)	industrial USB stick	consumer USB stick

All green: reuse without concern, viable even for production — provided capacity fits. Mostly yellow: backup role, secondary system. Red in more than one row: risk is usually higher than the savings.

A concrete example from the field

A customer brought us a Supermicro server from 2019: Xeon Silver 4114, 128 GB DDR4 ECC Reg, LSI 3008 in IT mode, 12x 3.5” bays, two redundant 750 W Platinum PSUs. The 12 old 4 TB nearline SAS disks had 42,000 operating hours. The server had been retired from a Hyper-V cluster.

Our recommendation: keep the server, replace the disks. The math:

Recycling variant:
  Chassis + CPU + RAM + HBA:  0 EUR (existing)
  12x new 12 TB enterprise HDD:  approx. 3,600 EUR
  2x NVMe for boot mirror:  approx. 120 EUR
  TrueNAS Community:  0 EUR
  Total:  approx. 3,720 EUR net, 96 TB raw

New-buy variant, same capacity:
  Xeon-D barebone 12-bay:  approx. 4,800 EUR
  12x 12 TB enterprise HDD:  approx. 3,600 EUR
  Boot NVMes:  approx. 120 EUR
  Total:  approx. 8,520 EUR net

The old disks went to electronic recycling — 42,000 hours and the remaining life is statistically no longer reliable for production data. The savings were around 4,800 euros, the server has been running for over a year as a secondary copy next to the primary TrueNAS and has proven itself in that role.

What you should never recycle

Two categories are ruled out for us on principle, even as test or backup systems:

• Hard drives with unknown history or above 50,000 operating hours. The price advantage is minimal, the failure risk real. A disk coming out of production and rotating in is not “used” — it is halfway to failure.
• Servers without working IPMI/BMC. Without a remote console, every TrueNAS maintenance turns into an on-site visit. At a secondary location that is simply not economically viable.

Conclusion: ask the right question

Recycle or buy new is not an ideological decision. It is a sober weighing of data role, operating cost and failure risk. An old Xeon with ECC RAM, HBA in IT mode and a Platinum PSU is often the right choice as a backup target — a consumer board with non-ECC RAM never is, no matter how cheap it appears.

DATAZONE supports you with exactly this evaluation: We look at your existing hardware, check compatibility with current TrueNAS SCALE versions and design a storage architecture with you that reuses existing investments where it makes sense and builds new where it must. Reach out via our contact form — we deliver an honest assessment instead of a sales pitch.