1. What "HDD not detected" actually means on screen

The same complaint covers at least five distinct states. Identifying the state first is what separates a ten-minute fix from a wasted service visit.

What Disk Manager shows	Meaning
No row at all	Drive not enumerated: cable, power, seating, or dead drive
Row present, health not Normal	Drive visible but failing or failed
Row present, capacity 0 or absurdly small	Unformatted, or filesystem not usable by the recorder
Properties show Read-Only	Detected and readable, deliberately or accidentally write-protected
Drive present and healthy, but no recording	Not a disk problem — schedule or record mode

The last row matters most. A healthy drive with no recording is a completely different investigation, covered by the three-layer recording model rather than by storage hardware. Treating it as a disk fault is how working drives get replaced and the real configuration problem survives untouched.

2. The check order that saves drives

Work down this list and stop at the first failure. The order is not arbitrary: each step assumes the previous one passed, so a physical fault is eliminated before any software action can destroy evidence.

Step one — power and cabling. The manual's installation guidance is direct: shut off power before replacing a drive, use a dedicated SATA drive for surveillance, and connect both the data and power cables to the drive and the mainboard. A drive that was inserted without the power lead seated will simply not appear, and no amount of formatting will change that.

Step two — detection. Open Main Menu > STORAGE > Disk Manager. If there is no row, the recorder never saw the device. Reseat, replace the SATA data cable, try another port, then try the drive in a PC. Only after a drive is proven good elsewhere should you suspect the recorder.

Step three — properties. If the row exists, check the Properties column. Read/Write is normal; Read-Only blocks all writing; Redundant has special meaning in a multi-drive unit.

Step four — format. If the drive is present and healthy but shows no usable capacity, format it. Formatting erases everything, so this is a decision, not a step to perform reflexively.

Step five — health. Only now look at health and SMART data. A drive that passes the first four steps but reports a degraded health state should be replaced, not nursed.

2.1 Tools, safety and the two minutes that prevent damage

The physical work is simple, and the mistakes are predictable. A short preparation routine removes most of them.

Power off and unplug. The manual instructs shutting off power before replacing a drive. Recorded footage is the reason: an interrupted write during removal can corrupt the filesystem index, turning a recoverable drive into one that needs a full format.

Photograph the bay before opening it. Cable routing differs between models, and a photo costs nothing. It also gives you a record of which drive occupied which slot, which matters when the interface's ordering does not match the chassis.

Handle by the edges, and keep the drive level. Mechanical drives tolerate a surprising amount of vibration while running and very little while being moved. Once removed, place them on a padded surface rather than stacking them.

Check the replacement before fitting it. Confirm the interface, the capacity and the class. Fitting an unsuitable drive and then discovering the mismatch after formatting wastes the visit and, on a live system, the footage.

Label the install date. A small sticker with the month and year on each drive turns future maintenance from archaeology into a scheduled task. It is the cheapest reliability improvement available on any recorder.

3. Disk Manager in detail

The Disk Manager page is where the recorder reports what it knows about attached storage. Four things live here, and each has a specific failure mode.

Information. Physical position, properties, status and capacity for each drive. On a multi-bay unit the physical position is the field that tells you which slot to open — worth checking before unscrewing anything, because drive order in the interface does not always match the order on the chassis.

Properties. Read and write, read-only, or redundant. Read-only is the setting that produces the most confusing support call in this entire area: the drive is detected, the capacity is displayed, old footage plays back, and nothing new is written. Nothing on screen looks wrong unless you read that one column.

Redundant. With two or more drives, one can be marked redundant so the recorder writes the same video to both. This is the practical alternative to RAID for many small sites: a live copy without rebuild complexity. Note the recorder's rule about ordering — designate the redundant drive in Disk Manager first, then enable redundancy in the recording schedule.

Format. Erases the drive. The dialog also offers to erase the HDD database, which additionally removes AI search data and uploaded audio files. On a recorder using AI features, that checkbox is a meaningful decision rather than a formality: clearing the database destroys the index that makes AI search fast, and it does not come back by itself.

3.1 Disk group mode versus quota mode

Multi-drive recorders can organise storage in two mutually exclusive ways, and the choice has permanent consequences for how capacity is allocated.

Disk group mode assigns specific drives to specific channels. A group of two drives might serve four cameras, and another group serves the rest. The advantage is isolation: a drive failure affects only the channels in that group, and write load is predictable. The disadvantage is inflexibility — spare capacity in one group cannot be used by channels in another, so a busy group can fill while an idle one has unused space.

Quota mode allocates a capacity slice per channel across the whole pool. Each channel has a target record duration and bitrate, and the recorder manages the space accordingly. This is the better fit for sites where every camera has a retention requirement rather than a hardware grouping.

The manual is explicit that the two cannot be active simultaneously, and that switching to quota mode may require formatting the disks. That single sentence is the reason to decide the mode before commissioning, not after: a mode switch on a live system can destroy the archive that the customer is relying on. If a site has been running in one mode for a year and someone "optimises" it into the other, the next support call will be about missing footage, not about capacity.

Practical guidance: choose disk group mode when the site has clearly separated functional zones with different criticality. Choose quota mode when the requirement is uniform retention across many cameras. Either way, record the choice in the handover pack, because it is invisible from the live view and expensive to rediscover.

4. Health: SMART and continuous monitoring

A drive can be detected, writable, and still be days away from failure. The recorder exposes two instruments for that.

Disk Check. The system detects HDD status so that failing drives can be replaced before they take footage with them. Results and SMART attributes are available per drive.

Health Monitoring. Under Storage > Disk Check > Health Monitoring, the recorder keeps a monitoring history. Select a check type and a time period, then search to see the recorded status over time. This is more useful than a single snapshot, because degradation is a trend: a drive that has been slowly accumulating reallocated sectors is a scheduled replacement, while one that jumped overnight is an emergency.

Two attributes deserve attention on surveillance drives. Read Error Rate climbing indicates the drive is struggling to retrieve data it previously wrote. Reallocated Sector Count increasing means the controller is retiring bad sectors — normal in small numbers over years, alarming when it moves quickly. Spin-up time drifting upward is the third signal worth watching on mechanical drives.

Health data is only useful if someone reads it. On managed sites, a monthly export of the disk check page alongside the recording verification is enough to catch most failures with weeks of warning.

4.1 Reading SMART numbers without panicking

Raw SMART values are easy to misread, and a technician who panics at the first non-zero number will replace healthy drives. Three attributes deserve context.

Read Error Rate is expressed as a large normalised number on most drives, and the raw value is not a count of failures. What matters is the trend in the normalised value and the threshold: as long as the value sits well above threshold and is stable across checks, the drive is doing its job. A falling normalised value with a rising raw value is the pattern to act on.

Reallocated Sector Count is the most honest indicator of physical wear. Every reallocated sector is a piece of the platter that has been retired because it stopped reliably storing data. A handful over several years is normal. A number that grows month over month means the surface is degrading, and the drive should be scheduled for replacement before it exhausts the spare pool.

Spin Up Time drifting upward is a mechanical warning. Bearings and motors wear; a drive that takes progressively longer to reach operating speed is telling you the end is closer than the capacity suggests.

Two reminders keep this in proportion. First, a single snapshot proves little — monitoring history is what turns attributes into a diagnosis. Second, health status "OK" is not a guarantee of remaining life; it means the drive has not yet crossed a threshold. That is why the pairing of health monitoring plus a scheduled replacement policy beats health monitoring alone.

5. Capacity: stop guessing, use the estimator

Most "the drive is too small" conversations can be settled with the recorder's own tool rather than with a spreadsheet argument.

Under Storage > Rec Estimate, the By Time tab takes a recording period and returns the required capacity; the By Space tab takes available capacity and returns how many days it will hold.

The manual is careful to label the estimate as reference data, and the caution is justified. Real retention depends on scene complexity — a busy street at night produces more data than a static corridor at the same nominal bitrate. When a customer's requirement is tight, measure: record for a full day, check the actual consumption, and extrapolate. That one measurement removes most of the uncertainty an estimate carries.

A second lever sits under Storage > Disk Quota, where capacity can be allocated per channel with a target record duration and bitrate. Quota is useful when one channel must not be starved by another — for example, a critical entrance camera alongside nine low-priority views. Note the manual's constraint: quota mode and disk group mode cannot be active simultaneously, and switching to quota mode may require formatting the disks, so plan the mode before commissioning rather than after.

5.1 A worked example you can reuse on site

Estimates are persuasive when they are shown as arithmetic rather than asserted. Here is a template that works for any site, with the numbers written out.

Take eight cameras recording continuously. Assume each main stream runs at 4 Mbps. One camera therefore writes 4 megabits per second, which is 0.5 megabytes per second, which is 43.2 gigabytes per day. Eight cameras produce roughly 345 GB per day before overhead.

Add ten to fifteen percent for container overhead, index data and filesystem slack, and the planning figure is about 380 GB per day. For thirty days of retention, that is roughly 11.4 TB. A 12 TB drive is marginal; a 14 TB drive leaves headroom for scene complexity and drive ageing.

Now change one variable at a time and watch the effect. Drop continuous recording to the sub stream at 1 Mbps and the daily figure falls to about 95 GB — the same drives now hold well over thirty days, but the continuous baseline is lower quality. Replace part of the continuous window with motion recording and the total depends entirely on how busy the scene is, which is why motion-only retention numbers should never be quoted without a measured day.

Two practical lessons come out of this. First, the difference between 4 Mbps and 2 Mbps is the difference between a 14 TB requirement and a 7 TB requirement — the bitrate decision matters more than the drive choice. Second, a customer asking for "thirty days at full quality on eight cameras" is asking for a specific capacity, and the honest answer is a number, not a reassurance.

The recorder's own estimator is the right cross-check. Enter the time period on the By Time tab, read the required capacity, and compare it with the arithmetic above. When the two agree, the design is sound. When they diverge significantly, the scene or the configuration differs from the assumption, and a measured day will tell you which.

6. RAID: powerful, and destructive if rushed

Multi-drive recorders support RAID, and the feature is genuinely valuable for sites that cannot lose footage. It is also the single most dangerous menu in storage, because creating a RAID group formats its member disks.

RAID type	Disks required	Notes
RAID 0	At least 2	Capacity and speed, no redundancy
RAID 1	Exactly 2	Mirror; half the capacity is protection
RAID 5	At least 3 (4–6 recommended)	Capacity of n−1 disks; common choice
RAID 6	At least 4	Two-disk tolerance
RAID 10	10	Mirroring plus striping

When you create a group, the recorder asks for a working mode that decides how resources are allocated between synchronisation and normal business. Self-adaptive adjusts synchronisation speed to current load, Sync first prioritises the rebuild, Business first prioritises recording and viewing, and Balance splits resources evenly. On a live site, Business first or Self-adaptive is usually the right choice; on a fresh install before cameras are attached, Sync first finishes the build quickly.

Automatic creation produces RAID 5 and only works when RAID 5 is the selected type. Manual creation walks through the type and disk selection. Either path formats the member disks.

A hot spare is the other half of a resilient design: an unassigned disk that the system can rebuild onto automatically when a member fails. A RAID array without a spare still requires someone to notice the failure and physically replace a drive — which, on an unmanned site, is exactly the delay the array was supposed to absorb.

6.1 What happens during a rebuild, and why it is a risk window

RAID is often described as if failure is invisible. In practice, a rebuild is a period of elevated risk, and technicians should be able to explain it to a customer.

When a member drive fails in a RAID 5 array, the array continues to serve data by reconstructing it from parity. That reconstruction is computationally expensive, and the remaining drives are now being read continuously at high load. A rebuild on a large array can run for many hours. During that window the array has no redundancy left: a second failure means data loss for the whole volume, and any latent sector errors on the surviving drives surface as read failures.

This is why two practices matter. First, a hot spare turns a manual replacement into an automatic one, shortening the exposure. Second, drive replacement should be scheduled before the array is under strain — a set of drives installed on the same day tends to reach end of life together, which is precisely the wrong pattern for a redundant array.

The recorder's working mode during synchronisation also affects the site. Business first keeps recording and viewing responsive while the rebuild proceeds slowly in the background; Sync first finishes faster but can degrade live viewing. On an active site, slow and stable is usually the better trade, and the rebuild will complete overnight anyway.

For sites that do not need RAID's capacity efficiency, the simpler redundant-drive arrangement described earlier — one drive designated redundant, redundancy enabled in the schedule — avoids rebuilds entirely. Two live copies, no parity mathematics, no risk window. It costs more capacity per stored hour, and it is dramatically easier to explain and support.

7. The ten causes, in check order

1. Data or power cable not seated. No row in Disk Manager. Most common physical cause on a new install.

2. Drive not supported or too small. Some models have minimum capacity expectations; check the model's documentation before blaming the drive.

3. New drive not formatted. Detected, unusable. Format on the recorder.

4. Properties set to Read-Only. Detected, plays back, writes nothing.

5. Drive failing or failed. Health not Normal, SMART attributes moving.

6. Drive is a PC-class unit. Works briefly, then degrades under continuous write duty.

7. RAID or disk-group mode change wiped the configuration. The drive is fine; the array was rebuilt and the data is gone.

8. Port or backplane fault. A different port on the same unit may detect the same drive.

9. Firmware or controller incompatibility. Rare but real on large drives in older recorders; check the compatibility note for the model.

10. It was never a disk problem. Channel offline, no schedule, or record mode not set — storage is healthy and idle.

8. Choosing the right drive

The manual's guidance is short and worth repeating: use an enterprise-grade or surveillance-grade drive, and PC drives are not recommended. The reason is duty cycle.

A desktop drive is designed for intermittent access — it spends most of its life idle and spins down. A surveillance recorder writes continuously, often from eight or sixteen streams at once, twenty-four hours a day, for years. The mechanical and firmware design differs: surveillance drives are built for sustained sequential write, tolerate higher temperatures in fanless enclosures, and handle the write patterns that recorder filesystems produce.

Capacity selection should follow the estimator, not the marketing instinct to buy the biggest drive available. Two mid-size drives with redundancy are usually a better design than one large drive with none, because a single large drive has a single point of failure and no rebuild path.

Finally, record the install date and model of each drive on the handover sheet. Surveillance drives are consumables on a multi-year horizon, and proactive replacement during a scheduled visit is far cheaper than an emergency call after footage is lost.

9. Commissioning checklist

#	Step	Pass condition
1	Power off before fitting the drive	No hot-plug on unsupported units
2	Both cables seated, drive screwed down	No movement when the unit is lifted
3	Disk Manager shows the drive	Row present with capacity
4	Properties = Read/Write	Not Read-Only
5	Format if new or suspect	Completes; capacity normal
6	Health status Normal	SMART clean
7	Rec Estimate for the requested retention	Capacity meets the requirement
8	RAID or redundancy configured if required	Status healthy, spare assigned
9	Recording schedule and mode verified	Playback shows a current band
10	Drive model and install date recorded	Handover pack updated

10. When the drive is fine and the recorder still records nothing

This is the most common misdiagnosis in the whole area. A healthy drive with no footage means the recording chain never delivered anything to write, and storage was never the fault.

Three checks resolve it quickly. Is the channel online? An offline channel produces no stream and therefore no recording, regardless of disk health. Is there a schedule with a band covering the required hours? An empty grid records nothing. Is the record mode set to Auto or Manual rather than Off? The mode overrides everything above it.

Those three questions, in that order, resolve the majority of "the drive isn't working" calls without touching Disk Manager at all. Our NVR recording troubleshooting guide walks that path in detail, including the disk-full strategy and the Motion & Alarm trap that silently changes which record types are written.

There is also a hardware-adjacent cause worth knowing: on PoE recorders, a camera that exceeds the port's power budget or the unit's total PoE budget will drop, and the recorder will appear to have "lost" a channel rather than a disk. If channels disappear in groups after a hot day or after adding a camera, check the power budget before opening the case.

10.1 Compatibility, large drives and firmware

Not every large drive works in every recorder, and the mismatch is rarely documented on the box. Three compatibility dimensions are worth checking before ordering hardware.

Capacity support. Older recorder models have per-drive capacity ceilings. A 16 TB drive in a unit that supports 8 TB may enumerate at the wrong size, fail to format, or appear intermittently. The model's specification sheet is the authority here, not the drive's marketing page.

Interface and sector size. Modern drives use 512e or 4Kn sectors. Most recorders handle 512e transparently and struggle with 4Kn. If a drive is detected but refuses to format, sector size is a plausible cause and worth checking before declaring the drive dead.

Firmware maturity. A recorder that has run for years without a firmware update may not recognise a current-generation drive. Updating firmware is a legitimate step — but it is also a change that can reset settings, so export the configuration first and schedule the update for a maintenance window rather than during a fault call.

A fourth dimension is power. Large drives draw more current at spin-up, and a multi-bay unit with an ageing power supply may fail to bring up a full complement of drives simultaneously. If drives are detected inconsistently — sometimes present, sometimes not — the power supply and the backplane deserve attention before the drives do.

The practical rule when replacing a drive: replace like with like where possible. Same class, same interface, same capacity band. It avoids compatibility questions entirely, and it keeps the array's behaviour predictable.

11. FAQ

Do I need to format a brand-new drive? Yes. It is detected but not writable until formatted. Formatting erases the drive, so confirm before proceeding.

Why does the drive show but recording never starts? Check Properties — Read-Only is the usual answer — then confirm the schedule and record mode.

Can I swap a drive with the power on? The manual says to shut off power before replacing a drive. Do not rely on hot-swap unless the model explicitly supports it.

Will formatting delete my AI search data? If you tick the option to erase the HDD database, yes — AI search data and uploaded audio files are removed. Leave it unticked unless you intend that.

How much capacity do I need? Use the Rec Estimate tool, then verify with a measured day of recording. Estimates are labelled as reference figures for a reason.

Is RAID worth it for a small site? Often a redundant second drive is a better trade than a RAID array: same protection against single-drive failure, far less operational complexity.

Why did a working system lose its footage after maintenance? A RAID or disk-group rebuild, or a mode switch that required formatting, is the likely cause. These operations format member disks by design.

Can I use a desktop drive to save money? The manual advises against it. Continuous write duty is a different workload, and the failure usually arrives when the footage matters.

11.1 Glossary of the terms on these screens

Disk Manager — the page that lists attached drives with properties, health and capacity, and where formatting happens.

Properties (Read/Write, Read-Only, Redundant) — the drive's role. Read/Write is normal operation, Read-Only blocks recording, Redundant marks a mirror target in a multi-drive unit.

Disk group — a set of drives dedicated to specific channels. Isolates failures, limits flexibility.

Disk quota — per-channel capacity allocation across the pool. Mutually exclusive with disk group mode.

SMART — the drive's self-monitoring data: error rates, reallocated sectors, spin-up time. Trend matters more than any single value.

Health Monitoring — the recorder's stored history of disk checks, used to spot degradation over time.

Rec Estimate — the recorder's capacity calculator, usable from either the time requirement or the available space.

Hot spare — an unassigned drive the array can rebuild onto automatically after a failure.

Redundancy (schedule setting) — writes the same video to a second drive. Backs up recordings, not snapshots.

Why does the recorder beep or show a disk alarm? Most units raise a disk exception when a drive fails, disappears or runs out of space. Silence the buzzer only after you know which of those three it is — the alarm is doing its job, and muting it without diagnosis converts a visible fault into a silent one that nobody discovers until footage is requested.

12. Sources

• Dahua Network Video Recorder User Manual V2.4.2 — §3.3 HDD Installation, §5.11 Storage, §5.11.2 Disk Manager, §5.11.3 RAID, §5.11.5 Disk Quota, §5.11.6 Disk Check, §5.11.7 Rec Estimate
• Dahua Technology — documentation portal

If you are sizing a recorder for a new site, the Dahua network camera range determines the bitrate side of the calculation, and the matching accessories line covers drives, mounts and power. Sites combining recorders with PTZ coverage will also want the PTZ collection for model matching.