What the message means
macOS had the drive, and then the drive stopped answering before macOS had finished putting it away. Bombich Software, who write the backup app Carbon Copy Cloner, describe the same thing: the drive drops off the bus for a moment, before macOS can cleanly unmount its volume.
Two words worth pinning down. A volume is the part of a drive you see on the desktop, with a name and files on it. To unmount a volume is to close it properly — macOS finishes writing anything still waiting, then lets go. That is what the alert is telling you did not happen.
The alert's second line is Apple's own instruction, and it is worth reading literally: Eject “name” before disconnecting or turning it off. The message is about the order of events, not about blame.
Here is the honest limit, and Bombich states it plainly: “macOS can't tell you why it dropped, only that it did.” Nothing on your Mac records the reason. What it does record is which device went, and when.
Find what dropped, and when
You have several drives attached, or the alert appeared while you were away and you missed it.
macOS keeps a running record of every drive appearing and disappearing. The part that does this is called diskarbitrationd — the background program that, in its own manual's words, “notifies clients of the appearance of disks and filesystems, and governs the mounting of filesystems”. The same framework puts the alert on screen, so its record is the one to read.
% log show --last 7d --predicate 'process == "diskarbitrationd"' --style compact | grep -E "created disk|mounted disk|ejected disk|removed disk"2026-09-10 01:06:51.015 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] created disk, id = /dev/disk6.
2026-09-10 01:06:51.017 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] created disk, id = /dev/disk6s2.
2026-09-10 01:06:51.018 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] created disk, id = /dev/disk6s1.
2026-09-10 01:06:51.388 Df diskarbitrationd[548:89c574b] [com.apple.DiskArbitration.diskarbitrationd:default] mounted disk, id = /dev/disk6s2, ongoing.
2026-09-10 01:06:51.405 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] mounted disk, id = /dev/disk6s2, success.
2026-09-10 01:07:06.337 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] unmounted disk, id = /dev/disk6s2, ongoing.
2026-09-10 01:07:06.382 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] unmounted disk, id = /dev/disk6s2, success.
2026-09-10 01:07:06.398 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] ejected disk, id = /dev/disk6, ongoing.
2026-09-10 01:07:06.405 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] removed disk, id = /dev/disk6.
2026-09-10 01:07:06.406 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] removed disk, id = /dev/disk6s2.
2026-09-10 01:07:06.406 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] removed disk, id = /dev/disk6s1.
2026-09-10 01:07:06.407 Df diskarbitrationd[548:88ed8eb] [com.apple.DiskArbitration.diskarbitrationd:default] ejected disk, id = /dev/disk6, success.Real output from our own Mac, unedited: all twelve lines this command returned for this one disk, from arriving to leaving. Other events in the seven days are not shown. It is a disk image being detached, not an external drive — we had no external drive to capture. Each line starts with the time and ends with the device, like /dev/disk6s2. “mounted disk” in the pattern also matches “unmounted disk”, which is why both appear. Follow /dev/disk6s2: mounted, then unmounted, then removed. /dev/disk6s1 is the instructive one — it was not mounted on this occasion, so it has no unmounted line, and it was still removed. Change 7d to cover when it happened; 24h or 1h reads faster. Our Mac's record reaches back about seven days, and anything older is gone.
The order is the useful part. Mounted means the volume was open and in use. Unmounted means macOS closed it properly. Removed means the device went away. For a volume that was mounted, a removal with no unmounted line before it means macOS did not get to close it first. That is what the alert describes.
One caveat, and the sample above shows it: a volume that was not mounted has nothing to unmount, so it is removed without an unmounted line and nothing is wrong. Check whether the volume was mounted before you read a missing line as trouble. We could not capture a real surprise removal to show you, because the Mac used for this page has no external drive and its record held no such event in seven days.
You get a device name, not the drive's name. To match /dev/disk6s2 to something you recognise, list what is attached now:
% diskutil listEvery disk and volume attached right now, with its name beside its device name. A drive that has already dropped will not be here until you reconnect it.
It disappears while you are away
The drive is there when you sit down, gone after lunch, and the alert was waiting for you.
A drive that is allowed to spin down can fail to come back, and macOS sees a device that stopped answering. Seagate reports the pattern in its own support note: “Most reports indicate that the issue is seen after the computer has gone to sleep.”
Apple documents the same shape on its own hardware. For the Mac Pro (2023), Apple says “certain models of internal SATA drives might unexpectedly disconnect from your computer after your Mac wakes from sleep”, fixed by updating macOS. That is a narrow case, but it shows sleep is a real cause and not a folk theory.
Fixin System Settings, Seagate advises turning off “Put hard disks to sleep when possible” and turning on “Prevent computer from sleeping automatically when display is off”. Your Mac will use a little more power. Give it a day before deciding whether it worked.
It happens now and then
An alert every week or two, on no particular schedule, sometimes when you nudge the desk.
Bombich puts the occasional case squarely on the connection: “If it happens once in a blue moon, it's almost always a loose cable or a sleepy USB port.” Seagate's own list agrees. It names a “Defective external drive cable”, a “Defective computer USB or Thunderbolt port” and a “Defective external desktop drive power supply”.
Fixchange one thing at a time. A different cable first, since that is the cheapest and the most likely. Then a different port. Then take any hub or dock out and connect the drive straight to the Mac. If the drive has its own power supply, try another one before you replace the drive.
The same drive, over and over
One particular drive drops repeatedly, on different cables and different ports.
When the connection has been ruled out, the remaining suspect is the drive or the case around it. Firmware is a small program built into a drive that runs the drive itself. Bombich names it as the repeat offender: “If it happens constantly on the same drive, the more likely culprit is the storage device's own firmware crashing and rebooting.”
Seagate lists “Failing external drive” among its causes too. Neither source treats a single alert as proof of a failing drive, and neither should you.
Fixcopy anything you care about off the drive first, while it is still readable. Then open Disk Utility, select the drive and run First Aid, which checks the volume and repairs what it can.
If it still drops after that, Bombich points at the drive. His words: “the problem may be inside the drive itself, in which case erasing it -- or ultimately replacing it -- is the path forward”.
Erasing destroys everything on the drive and cannot be undone, so it only comes after your files are somewhere else. A drive that behaves like this should not be the only copy of anything.
Ejecting, and why it matters
Ejecting is not ceremony. The manual for diskutil, the Mac's own disk tool, says an eject makes removable media “eligible for safe manual removal”, and that “the disk's volumes will first be unmounted”. The unmount is the part that matters, and it is what gets skipped when a drive vanishes.
The same manual notes an unmount can take “up to a few seconds (or more)”, because macOS waits for anything still using the drive and finishes writing. That wait is the reason a drive pulled the moment you drag it to the Trash can still complain.
Fixeject from Finder and wait for the drive to leave the desktop before you unplug it, turn it off, or undock the Mac.
What Noah can check on this Mac
NoahNoah can read the record described above and name the device that dropped, along with the time it happened. It can also list the volumes mounted right now, and how full each one is.
Noah cannot tell you why the drive dropped — no Mac can, and that limit is the honest one. That list of mounted volumes will not show a drive that is attached but has nothing mounted, which is the case this page is about. It cannot match a device name back to the drive's own name once the drive has gone. It cannot check a drive's health, or the state of a cable. It does not repair hardware, and it does not recover data from a failing drive. It also cannot see an event older than your Mac's record, which runs to roughly a week.
The short version
- The alert means a drive stopped answering before macOS finished closing it.
- It names the volume. The log adds the device name and the exact time.
- A clean removal shows unmounted, then ejected, then removed.
- Occasional alerts point at a cable, a port, a hub, or sleep.
- The same drive dropping constantly points at the drive or its case.
- Copy your files off before running First Aid, and long before erasing anything.
- macOS records that it happened, never why.