TS-253 Pro Very High HDD Wake-ups

[Nicanor]

Hi,

I have this nice NAS for about 3-4 days and after the purchase i only updated to the latest version of firmware from this month (4.1.3 build 0408) and configured it.
The problem is that i know that some WD Red HDDs have a LCC problem i started to monitor the S.M.A.R.T. daily and i found something that scared me, but not something related to that LCC problem. The problem is in fact that the NAS is turning on the HDD almost 30 times a day. In almost 3 days, the start stop count was almost 70.
I am the only user and i leave it almost all day in standby.
Below i will explain what settings i have, maybe it helps:
1. I configured myCloud service and also the upnp function and the ports are automatically port forwarded (i know this is not the best way, but it's just for testing now). This also enabled the QNAP DDNS that in settings shows that i automatically refresh from 15 to 15 minutes (i don't know if this wakes the HDD also, i don't think so)
2. I added Photo, Video, Audio stations but then disabled them.
3. I also installed the hybrid station for transconding codecs, but i also disabled this in order to find what is causing these wake ups.
4. I checked the crontab and there where like 10-15 things but almost all of them where supposed to run at night. Firstly i tried to make them all run at almost the same hour in order to wake the HDD only then, but with no luck, then i put # in front of almost all and still the HDD wakes up a lot. After that i restarted the crond service and the whole NAS and still the HDD still wake up.
5. I had dlna media server and twonky enabled by default and i disabled the internal one and only left Twonky. I even updatd the Twonky server with the one from App Store, version 8. After that i tried to set the media server to scan when file system changes (with -1) .. and also to disable scanning (with 0). Still the same, the HDD wake up a lot.
6. I have wake on lan anctive, can this be a cause, the router or something to wake the HDD ?
7. I followed the links from similar issues from this forum and put fixed IP to the NAS.
8. I have HDD standby set to 15 minutes (i had it also on 30) and the NAS enters standby as it should, but wakes up a lot.

After that i tried also the tool to monitor disc activity (blkdevMonitor.sh) and after some loops (5-10) it stops logging, and then after an hour or so it starts, logs 2 loops and hangs again. After that after 2 hours (at precisely the same minute) it wakes up again and logs 2 loops.
The loops are very short and contain only 8 lines of the following:
<4>[94606.343930] vgdisplay(24621): READ block 8 on md2 (8 sectors)

I don't have the full log because i lost it after restart (used temp folder for script), but if the full log helps i will run it in root to persist.

So after these wake ups, only the type of record as above is shown.

I don't know what to do anymore to keep the HDD in standby, because 30 start/stop per day is too much and sooner or later my HDD will fail.

Can someone help me with this, because i don't know what to do anymore.

[schumaku]

I don't know what to do anymore to keep the HDD in standby, because 30 start/stop per day is too much and sooner or later my HDD will fail.

That's hype nonsense. If you are concerned about lifetime, keep the HDD up and running 24*7.

[Nicanor]

I know that may be a possibility, but it's not normal.
I want to find the cause of this because it's something strange, none of the programs is to blame, so may be a bug or something in the firmware and maybe somebody from QNAP could find it if it is indeed a bug.
I say this because all the lines from that log say something like this (8 identical lines, each time):
<4>[94606.343930] vgdisplay(24621): READ block 8 on md2 (8 sectors)

So i found the this is indeed an OS command, so who may call it and why ? It is regularly and 1h then at 2 .. etc .. and it's not something from cron, maybe there is a different hidden cron that i don't know about ?

[schumaku]

The HDD do spin-up when required. Can't see anything abnormal.

How many times is vgdisplay (yes, it's an LVM2 command) called? Any relation to open QTS desktop sessions?

[Briain]

Each WD Red takes less than 3W more when spinning (then when hibernated), so for a two bay NAS with WD Red disks, IMHO if there is any sign of multiple spin up/down cycles, then it makes no sense to even think about attempting to get disk hibernation working; even under ideal conditions, the annual savings would be far less than it costs for two pints of fizzy lager in a city bar (and in reality, probably more like one pint).

That's one of the key reasons why with my latest Qnap, I've moved from using enterprise disks (which I always had used, in the past) to now using WD Red disks; they draw so little additional power, that I can just leave them spinning 24/7 (and I no longer have to worry about checking that the disks aren't spinning up and down many, many times per day, thus temperature cycling them many, many times per day; they just keep spinning 24/7 and thus everything sits at a nice, constant temperature and IMHO, that's the ideal environment for electromechanical devices).

Bri

[Nicanor]

@Briain Thanks. It's not about the power consumption. If the HDD is consuming 3W more in spinning mode that hibernate, then the annual consumption will be like 26-27 kw .. so it's not a big deal. What i'm woried about is the wearing of the mechanism. Is 24/7 working WD Red safe from wearing point of view ? What is worse, to keep it always spinning or to have many start stop ?

@schumaku I know that it's working as it should but the problem is that is spinning up and then stays in idle and goes again into hibernate. So it's not doing something useful. Regarding the vgdisplay command it's the only thing that appears in those logs when the HDD exist hibernate. It last for about 2-3 loops, with exactly 8 vgdisplay commands in log, nothing else, with the first 7 having the same PID and the last with a different one, then in the log there is a pause again for exactly 1h. Then again it wakes up, 2-3 loops and sleep. The after exactly 2 hours it starts again.
It's a new NAS and i don't have anything on it besides the samples and i checked from work with ssh the status of the HDD and also the logs. So there is NO QTS desktop session active when the HDD is waken up.
Maybe that log is not logging the exact cause of the HDD wake up. Do you know some other way to monitor this ?

[Briain]

@Briain Thanks. It's not about the power consumption. If the HDD is consuming 3W more in spinning mode that hibernate, then the annual consumption will be like 26-27 kw .. so it's not a big deal. What i'm woried about is the wearing of the mechanism. Is 24/7 working WD Red safe from wearing point of view ? What is worse, to keep it always spinning or to have many start stop ?

Hi

I used to be of that mind, but when you think about it, the weekly full SMART test (assuming you have elected to set that, and if not, it is well worth considering doing so) keeps them spinning over one night, they get spun up every night at 03:00 for scheduled maintenance, they get spun up whenever I play a tune on my HiFi and now that all my data is on my NAS (I've nothing on laptops, desktop or tablets, other than the files/documents I'm actually working on) they get spun up several times a day for legitimate access reasons. Add to that the possibility of unscheduled spin up events due to some service or package on the NAS (for a while Twonky used to be bad for that - fine after a reboot, then after a few days, there were about 4 spin up events per day - but I'm not sure what the latest versions are like) and that I think some other of other network devices (media boxes and the likes) were causing spin up events, and you end up with it clocking huge amounts of start stop counts per week (and a lot of hassle trying to limit them).

Agreed about the power not being a big deal, and of course, in reality it's even less than that, when you think about it. Given the above, it means the yearly saving is far less than 26 KWh/y as even with hibernation enabled, as they'll still spend quite a lot of time spinning. So that leaves the wear issue and my concern is that assuming hibernation will never be perfect, it means they'll be cycling temperature every time they spin up or stop and cool down (probably about 10 degrees, so admittedly not at all bad for a WD RED) so there's will be expansion and contraction of motor, platter bearings and the platter, itself. I guess wear might also be enhanced by the rapid spin up torque loadings to the motor and platter bearings, so given all of the above (and assuming multiple spin up events per day), it makes me wonder if wear might be less severe if they're just left spinning and comfortably sitting at a constant temperature (so no expansion and contraction cycles). I also prefer to keep them spinning as it removes the inrush surge loading a spin up event creates (so less work for the power supply and motor switching devices on the disk control board). Basically, I generally like the idea of things sitting at equilibrium, not cycling.

On balance, I decided the wear would possibly be less by adopting the continuous spinning option, but as alluded to above, that's because I found it difficult to get reliable hibernation in my netwoek and with my usage patterns (I did try, even to the extent of running two VLANs to keep the NAS network 'quiet'. At the end of the day, I decided life would be less hassle by retiring my TS-659 (RAID 6 enterprise disks) to being an occasionally on backup server and getting a TS-453 with a couple of big WD Reds in RAID 1 (with slots for future expansion) and leaving that spinning. Even spun up, it's way less than the TS-659 was hibernated, and now that I'm running a Windows 7 VM on it, goodness knows how well hibernation would have worked, anyway (actually, I must test that to see if the Win 7 VM does let it hibernate).

With the TS-659, I used to run scripts to schedule hibernation at night and continuous spin through the day, as well as weekly writing down spin up count from SMART (to ensure it wasn't cycling many times a day); now, life is much easier as I don't have to bother with any of that, nor concern myself that every time I wanted a piece of information from my NAS (outside its normally spinning times of day) I'd clock uo yet another spin up and temperature cycle event; now, life is so much less hassle!

Bri