Does this HD sound normal?

[Johhhn]

I am replacing my 3TB reds with 5TB Toshibas. I noticed last night that my Toshiba is making a disk access 'sound' every few seconds. It doesn't make this sound if I'm in the BIOS or while it's loading up FreeNAS. As soon as it mounts the drives, it starts to do this. The drive is louder than the WDs and is reminiscent of how old drives can be heard much more easily when the disk is being accessed. Is it normal for the drive to be accessed every few seconds, if this is what the noise is about?

SMART looks good, nothing strange at all, fyi.

http://tinypic.com/r/s48fah/8

 

[JDCynical]

That makes me nervous, and I'd source a replacement as soon as I could. SMART is far from infallible, and I've seen drives that SMART didn't bat an eye on... until after it completely failed.

Of course, I'm also a bit paranoid after being bitten by SMART missing an impending drive failure, so YMMV.

EDIT: Does dmesg show anything out of place? I had a drive failure not too long ago that dmesg was showing there were problems (timeouts in this case) well before SMART said things were not five by five.

 

[cyberjock]

There isn't enough time video to say for 100% certainty. Generally when you do work on the drive itself, the seeks are random. If you have a clickety-clickety that is at a regular recurring interval and it only happens when the system is in use, then you may have a problem. If it is irregular or it happens when there is zero disk activity (you must verify there is zero disk activity) then the disk is probably fine.

To be honest, the chances of a disk being problematic despite SMART saying they are fine are extremely remote. Yes, it can happen, but it is so incredibly rare that you shouldn't bet money (or your data) on it. After all, the whole purpose of redundancy is to be ahead of failures. If someone was to say "well, I don't have enough redundancy to handle this if it is failing" then you have bigger problems because you obviously haven't adequately prepared and execute a plan to protect your data sufficiently.

 

[joeschmuck]

What is the model of the Toshiba hard drives you are using? Have you done any research on them before purchasing them such as some solid reviews to see what their characteristics are?

EDIT: Do all these drives make the noise or only one drive?

 

[DrKK]

I have had *TWO* guys in the IRC channel ask this same exact thing, each of them with 5TB Toshiba's. I don't know if the OP is related to either/both of the guys I saw in IRC, but there appears to be evidence that the 5TB Toshiba has a click click sound, occurring every few seconds, and it is characterized, to my surprise, as "quite annoying" and "audible over fan sounds across the room". One of the guys in IRC said the click sound was annoying enough that he was actually not satisfied with the drives.

So this may be a thing. If so, though, I'd expect there'd be all kinds of hemming and hawing on forums Out There in the Real World...

 

[Johhhn]

Nope, don't know those guys and I haven't been on IRC in a long time.

I tested two new 5TBs to see if they made a similar sound and all of them do this when I run a SMART Short Self-Test.

I think the drives are probably fine and may be 'normal'.

http://tinypic.com/r/9tk5cm/8

Joe: these are 5TB MD04ACA500. I've read the drives are noisy, like old school drives. but some say they are quiet. I have two of the 5TBs installed, but can't tell if both of them make the noise (can't isolate the sound with 6 drives installed.) And the noise only occurs after FreeNAS mounts the drives.

I'm going to replace the suspect 5TB with another and see if it does the same thing.

I have had *TWO* guys in the IRC channel ask this same exact thing, each of them with 5TB Toshiba's. I don't know if the OP is related to either/both of the guys I saw in IRC, but there appears to be evidence that the 5TB Toshiba has a click click sound, occurring every few seconds, and it is characterized, to my surprise, as "quite annoying" and "audible over fan sounds across the room". One of the guys in IRC said the click sound was annoying enough that he was actually not satisfied with the drives.

So this may be a thing. If so, though, I'd expect there'd be all kinds of hemming and hawing on forums Out There in the Real World...

 

[joeschmuck]

Do me a favor, post the SMART data for the drive, "smartclt -x /dev/ada0" for example, yes -x switch please. Then after 10 minutes or so, a second set of values for the drives. Do this for both Toshiba drives.

Also, post a screen capture of the "View Disks" GUI screen ensuring all the values are displayed for the Toshiba drives.

I want to ensure you are not having a drive failure and you don't have something setup causing it. It's the only way I know to give you proper feedback. And it may be normal for these drives and sounds like the head actuator is causing the noise, but lets just see. I'll bet it get much noisier when you start reading random data.

Also, where is your system dataset being stored? Likely on your pool vice the boot device. And what do you have for plugins running?

 

[joeschmuck]

Although it looks like your Toshiba drives do not support AAM, you should still give it a try. A value of 128 may work and is typically considered the most quiet setting. The other option may be to test out APM settings however only test one drive and open up the case so you can actually hear the drive you are testing. Again, with APM try a value of 128. If you go below 128 the drive should spin down however depending on activity it will likely spin right back up. Going above 128 may help as hese values typically adjust head parking delays. For all of these settings, I could not find anything specific for your drive so who knows if thee will work, you will need to test it out.

While testing, ensure you have your plugins (if any) turned off and turn off CIFS and NFS, well all sharing activities. You will want to isolate your system the best you can in order to figure this out.

 

[Johhhn]

While testing, ensure you have your plugins (if any) turned off and turn off CIFS and NFS, well all sharing activities. You will want to isolate your system the best you can in order to figure this out.

Ok, I did some more extensive testing. The drive access is so much louder than my WDs that I didn't think it could just be *normal*. So, I did some extensive troubleshooting based on some stuff you mentioned and discovered that it writes to the disks every few seconds and these were the disk activities I discovered:

1. When you are logged into Web GUI:
   1. 1117133511464 80 3437 nginx vop_write 232B /mnt/huge/.system/syslog/log/nginx-access.log
2. Also,Plex
   1. 367227635024 972 6780 Plex Media Serve vop_write 141B /mnt/huge/jails/plexmediaserver_1/media/<unknown>

• And a Jail i had setup for backups for a vps

1. 366561034373 0 3480 collectd vop_write 4K /var/db/collectd/rrd/localhost/df-mnt-huge-jails-vps-backups-sftp/df_complex-free.rrd

When I shutdown those 3 things, there is no disk activity (writes) every few seconds and it's much quieter.

So guess these drives are OK?

If you think I should still do what you suggested with the smartctl -x, let me know!

 

[DrKK]

The smartctl -x readouts will give us the opportunity, as guys that have looked at thousands of these, to just take a look at all of the stats and make sure something subtle, or not-so-subtle is going on. It's a free service from subject matter experts who do this all week long. I suggest you take advantage of it. :) It's about 250x more skill than the Geek Squad, and the price is $0.

do this, and pastebin (or similar) the results for us:

smartctl -x /dev/adaN

where N is the numeric identifier of the Toshiba drives. You can tell what the identifiers are for the Toshibas by a "camcontrol devlist".

 

[joeschmuck]

I think you are fine. Those drives are just loud and I've seen a few postings on the internet suggesting as such, however not too loud. You can try the AAM and APM settings in an effort to see if you can mitigate some of the noise if it bugs you but I suspect you will have to live with it.

 

[Johhhn]

Here are the links of the output. These drives (7200rpm) definitely run hotter. Need to add more fans now.
All three Toshibas that I currently have installed. Fresh reboot. Disk being accessed every few seconds as in video.
http://pastebin.com/Zx3sXc2t

After 1o mins
http://pastebin.com/WrnfK2J7

Thanks for the help guys!!

The smartctl -x readouts will give us the opportunity, as guys that have looked at thousands of these, to just take a look at all of the stats and make sure something subtle, or not-so-subtle is going on. It's a free service from subject matter experts who do this all week long. I suggest you take advantage of it. :) It's about 250x more skill than the Geek Squad, and the price is $0.

do this, and pastebin (or similar) the results for us:

smartctl -x /dev/adaN

where N is the numeric identifier of the Toshiba drives. You can tell what the identifiers are for the Toshibas by a "camcontrol devlist".

 

[joeschmuck]

Your drives look good. One thing I noticed which you may not have is these drives leave the heads loaded the entire time. I'm not sure if that is because you may have routine activity however I don't think that is the case so you should be aware that your hard drive heads do not park. I'm not insinuating there is a problem, my drives don't park either which is my choice.

 

[Johhhn]

Thanks for looking, Joe! I feel better about these drives.
I don't like it when drive park. I feel much better leaving things spinning 24/7. Historically, for me, I've had drives last MUCH longer when on 24/7 and never going to sleep/parking.

Thanks again! Going to add the other 3 now.

Your drives look good. One thing I noticed which you may not have is these drives leave the heads loaded the entire time. I'm not sure if that is because you may have routine activity however I don't think that is the case so you should be aware that your hard drive heads do not park. I'm not insinuating there is a problem, my drives don't park either which is my choice.

 

[Bidule0hm]

Spin down and parking are two very different things. Parking is just that the heads are off loaded of the platters to avoid useless wear.

 

[Johhhn]

It could be the disk access every few seconds that's happening from Plex.

How many of you have your drives parking?

I'm not sure if that is because you may have routine activity however I don't think that is the case so you should be aware that your hard drive heads do not park. I'm not insinuating there is a problem, my drives don't park either which is my choice.

 

[joeschmuck]

Looking at your drive information shows that the heads have loaded the same number of times at power was applied. The only way to know for certain if they will automatically park after some unused period of time you could take one of your new drives which you have not yet installed, connect it to FreeNAS hut do not add it to a pool or anything, check the smart data (IDs 4, 12, and 193). 4 = Spin up/Spin down times, 12 = Power applied, 193 = Heads loaded. Wait at least 30 minutes or more and run a smart short test to wake up the drive and then check smart data again to see if the IDs are not equal. If they are equal then wait some more time and try all over again. If after 1 hour you cannot get the IDs to be different then I'd have to say the heads will never park due to disk inactivity. After thinking about it some more, this is not a NAS drive, it's a desktop drive right? So the heads should park after some period of inactivity.

And like @Bidule0hm said, spinning down is not the same as head parking however I suspect you know that since you related the two events.

 

[Johhhn]

I did it over half hour and a couple of hours and the only thing that changed between the 4,12, and 193 was the LCC. It was 1, then 4, and now 9.

Interestingly enough, the 9 is higher than the other drives that have 79 power on hours vs the 4 this new one has.

Looking at your drive information shows that the heads have loaded the same number of times at power was applied. The only way to know for certain if they will automatically park after some unused period of time you could take one of your new drives which you have not yet installed, connect it to FreeNAS hut do not add it to a pool or anything, check the smart data (IDs 4, 12, and 193). 4 = Spin up/Spin down times, 12 = Power applied, 193 = Heads loaded. Wait at least 30 minutes or more and run a smart short test to wake up the drive and then check smart data again to see if the IDs are not equal. If they are equal then wait some more time and try all over again. If after 1 hour you cannot get the IDs to be different then I'd have to say the heads will never park due to disk inactivity. After thinking about it some more, this is not a NAS drive, it's a desktop drive right? So the heads should park after some period of inactivity.

And like @Bidule0hm said, spinning down is not the same as head parking however I suspect you know that since you related the two events.

 

[joeschmuck]

So that tells me that the heads will park if given enough inactive time. I guess your FreeNAS pool is active more frequently that that parking limit which in my opinion is a good thing for you. You never want to park the heads and a few seconds or minutes later load them again, rinse and repeat.

If your drives can sleep (should you ever want it), you would need to move your jails and system data set to a different device like a SSD. When FreeNAS 10 comes out I have every intention of testing that out to see if I can significantly lower my power draw. I may not stay like that but I will at least test it out. I know I've done these tests before a few years ago and the results are somewhere in these threads but I have made CPU power saving tweaks so there should be a difference.