I have a WD Elements 25A3 - a 16 TB external SATA drive. It's (I think) a little over a year old.
Formatted NTFS on Windows 10.
I am not having any problem accessing files on the drive.

However, recently it has started making a soft but audible knocking/clicking sound every 5 seconds. The 5 second interval is very consistent. I listened and timed it closely over a minute or two and once it seemed to miss a cycle and go 10 seconds until the next knock/click but then it resumed the next 5-second cycle. If I put my hand on the drive, I can feel something kind of bump when I hear the knock/click.

In Windows Explorer, I right-clicked on Properties for the drive and went to Tools / Check. It said:
You don't need to scan this drive
We haven't found any errors on this drive. You can still scan the drive for errors if you want.

I did scan the drive and it found no errors.

If I use the Windows Safe-Remove Hardware and tell it to remove the drive, the knocking stops. But if I power cycle the drive to make Windows redetect it, the knocking starts again.

Should I be worried it is on the verge of failing?
What should I do now?

Examine the SMART report with a tool such as CrystalDiskInfo or GSmartControl.

_________________
A backup a day keeps DR away.

Thank you for the reply and the suggestion!

I downloaded GSmartControl and executed the short (2-min) self test. It passed without any errors.

I'm not seeing any other problems, though this is the first time I've used the tool so I may be missing something to look for.
But e.g.:

In the General tab, the Overall Health Self-Assessment test PASSED.

In the Attributes tab, there are a series of SMART Attributes data structure. For all the IDs, the Failed column says "never".

In the Statistics tab, it includes:
General Error Statistics (rev 1),
with the following 2 lines showing 0 values for errors/resets.

In the Error Log tab, it says:
Complete error log:
SMART Extended Comprehensive Error Log Version: 1 (1 sectors)
No Errors Logged

In the Temperature log, it includes the line:
SMART Status: 0xc24f (PASSED)

It looks like the only other thing I can do with this tool would be to run the Extended self-test, which it says will take 33 hrs.

You don't have to worry, if You have backup.

I'm not sure if this reply was tongue-in-cheek, but I don't consider this to be helpful. We're always supposed to have a backup, right? The reply doesn't answer any questions about this. Such as, whether this hard drive has a much more likely probability that it will fail soon compared to any other drive, versus, some drives just do this without it being the precursor to a failure. Or if there is anything that may make the noise go away (is it something that can physically be repaired, is there some software setting that may make a difference, etc). Does anyone even know if there is some significance to the 5-second period? The drive is probably spinning many times per second, so what is it that would happen every 5 seconds? So, lots of questions. I was hoping someone here may have something more than this answer.

That's not useful. The most important data are the raw values of each attribute. A hard drive can have thousands of bad sectors and still "pass" the tests.

_________________
A backup a day keeps DR away.

I'm sorry I didn't know to include all those, but the only direction I had was "Examine the SMART report". I had no idea what was and was not significant in all the various tabs, and I was afraid of giving information overload.

Here are all the attribute values:

"ID","Name","Failed","Norm-ed value","Worst","Threshold","Raw value","Type","Flags"
1,"Raw Read Error Rate","never","100","100","1","0","<b>pre-failure</b>","PO-R--"
2,"Throughput Performance","never","134","134","54","104","old age","--S---"
3,"Spin-Up Time","never","84","84","1","298 (Average 354)","<b>pre-failure</b>","POS---"
4,"Start / Stop Count","never","95","95","0","2351","old age","-O--C-"
5,"Reallocated Sector Count","never","100","100","1","0","<b>pre-failure</b>","PO--CK"
7,"Seek Error Rate","never","100","100","1","0","old age","-O-R--"
8,"Seek Time Performance","never","128","128","20","18","old age","--S---"
9,"Power-On Time","never","99","99","0","8815","old age","-O--C-"
10,"Spin-Up Retry Count","never","100","100","1","0","old age","-O--C-"
12,"Power Cycle Count","never","66","66","0","2351","old age","-O--CK"
22,"Unknown Attribute","never","100","100","25","100","<b>pre-failure</b>","PO---K"
192,"Head Retract Cycle Count","never","100","100","0","2587","old age","-O--CK"
193,"Load / Unload Cycle","never","100","100","0","2587","old age","-O--C-"
194,"Temperature (Celsius)","never","30","30","0","46 (Min/Max 16/53)","old age","-O----"
196,"Reallocation Event Count","never","100","100","0","0","old age","-O--CK"
197,"Current Pending Sector Count","never","100","100","0","0","old age","-O---K"
198,"Offline Uncorrectable","never","100","100","0","0","old age","---R--"
199,"UDMA CRC Error Count","never","100","100","0","0","old age","-O-R--"

SMART is OK. The drive is not reporting any bad or weak sectors.

One surprising thing I notice is the Power Cycle Count:


This would suggest that the drive is rated for only ~7000 power cycles ???

What is the model number of the drive inside the enclosure, as identified by your SMART tool?

_________________
A backup a day keeps DR away.

check for your power adapters. Drives do make clicking sound when they are deprived of power.

--

_________________
THERE IS ALWAYS HOPE TILL YOU LOSE IT.

Background scanning? Patrol seeks?

_________________
A backup a day keeps DR away.