Hard drive reliability

[Per Hansson]

Re: Hard drive reliability

This is in my parents Firewall, Proliant 800 server with dual Pentium Pro 200Mhz CPU's running 24/7.

Host adapter information:
ncr53c8xx-3.4.3b-20010512

Drive information:

/dev/sg0 [=/dev/sda bus0 ch=0 target=0 lun=0]
    /dev/sg0
    Vendor = COMPAQ
    Product = ST32171W
    Version = 0388
    Serial Number = JE917993
    Copyright = Copyright (c) 1996 Seagate All rights reserved
    SCSI Firmware = 83450388
    Servo RAM Release = 83466246
    Servo ROM Release = 83467326
    Servo RAM Date = 1209
    Servo ROM Date = 0110

    Blocksize = 512, Highblock = 4109999, Capacity = 2055 MB
    -this is a Seagate drive
    -this drive does not support DST
    -Mode Page Settings [current value (default)]:
        -WCE bit = 0 (0)
        -RCD bit = 0 (0)
        -AWRE bit = 1 (0)
        -ARRE bit = 1 (0)
        -DExcpt bit = 0 (1)
        -Number of cache segments = 3 (3)

AFAIK it does not even support SMART stats, do I win a price?

[goontron]

Re: Hard drive reliability

^ Yes, now that you have posted this the drive will crash.

[mariushm]

Re: Hard drive reliability

Modern hard drives seem to be more reliable.

Anyway... time to buy another 4 TB drive to replace the 1 TB drive....




I still have a Fujitsu 350 MB SCSI drive somewhere ... it probably works, but I'd have to find a computer with an ISA slot to fit the Adaptec SCSI controller to be sure..

[stj]

Re: Hard drive reliability

fujitsu scsi? bound to work.
fujitsu & ibm scsi drives were the best i ever had.

[eccerr0r]

Re: Hard drive reliability

BTW that Seagate 750GB PATA I have is a 7200.10 and it has 8 remaps on it after just a bit under 20K POH... Hasn't grown in a while and haven't lost data from it, but it wouldn't be placed under the same category as my 120G disks.

I don't think any of my 500G disks have remaps=0... well maybe 1 or 2.

[delaware74b]

Re: Hard drive reliability

I have a WD Black 640GB in my desktop that I use as a server/audio streaming uploader. It has been running 24/7 for the past 4 years since my move to Dover. Sitting on bench running 30C in basement - 64K hours with less than 200 power on cycles..... no pending sector reallocation or any errors.

[kc8adu]

Re: Hard drive reliability

from my main desktop.
its os only.
storage is on a raid 5.
34k on it.
had over 100 these as server pulls.
its a velociraptor marked enterprise storage.

[goontron]

Re: Hard drive reliability

Heres more of a WTF than a reliability comparison....

[Heihachi_73]

Re: Hard drive reliability

Seagate ST340015A, 40GB IDE. Brand new in 2005, still running Windows XP 24/7 in a 2.8 GHz Celeron*. 17865h = 96357 power on hours.
Seagate ST360015A, 60GB IDE. Found in hard waste in 2006, still running 24/7 in same PC as above. 1504Eh = 86094 power on hours.

*The exact same PC from my very first post on this site. The one and only thing that has ever gone wrong with that PC started with 'T' and ended with 'po'. In case you're wondering, that PC is still using the same $2 Thermal Master PSU I mentioned in that post!

[eccerr0r]

Re: Hard drive reliability

I've seen quite a few drives with huge head load counts. Green WD's and laptop disks gets the worst. I stopped my 2TB wd green from unloading but not in time before it got several hundred thousand:

 1 Raw_Read_Error_Rate   0x002f  200  200  051  Pre-fail Always    -    0
 3 Spin_Up_Time      0x0027  178  168  021  Pre-fail Always    -    6075
 4 Start_Stop_Count    0x0032  100  100  000  Old_age  Always    -    189
 5 Reallocated_Sector_Ct  0x0033  200  200  140  Pre-fail Always    -    0
 7 Seek_Error_Rate     0x002e  200  200  000  Old_age  Always    -    0
 9 Power_On_Hours     0x0032  046  046  000  Old_age  Always    -    40045
 10 Spin_Retry_Count    0x0032  100  100  000  Old_age  Always    -    0
 11 Calibration_Retry_Count 0x0032  100  100  000  Old_age  Always    -    0
 12 Power_Cycle_Count    0x0032  100  100  000  Old_age  Always    -    186
192 Power-Off_Retract_Count 0x0032  200  200  000  Old_age  Always    -    135
[COLOR="Red"]193 Load_Cycle_Count    0x0032  001  001  000  Old_age  Always    -    624808[/COLOR]
194 Temperature_Celsius   0x0022  112  097  000  Old_age  Always    -    38
196 Reallocated_Event_Count 0x0032  200  200  000  Old_age  Always    -    0
197 Current_Pending_Sector 0x0032  200  200  000  Old_age  Always    -    0
198 Offline_Uncorrectable  0x0030  200  200  000  Old_age  Offline   -    0
199 UDMA_CRC_Error_Count  0x0032  200  198  000  Old_age  Always    -    1819
200 Multi_Zone_Error_Rate  0x0008  200  200  000  Old_age  Offline   -    0

Nowhere near that 9 million but this is also pretty darn bad that WD thinks that this drive is just about done with head loads (since it's down to 1).

[goontron]

Re: Hard drive reliability

^ Wish i could figure out how to stop it... hdparm can't seem to trim the load-cycles on this drive.

[kevin!]

Re: Hard drive reliability

Seagate ST340015A, 40GB IDE. Brand new in 2005, still running Windows XP 24/7 in a 2.8 GHz Celeron*. 17865h = 96357 power on hours.
Seagate ST360015A, 60GB IDE. Found in hard waste in 2006, still running 24/7 in same PC as above. 1504Eh = 86094 power on hours.

*The exact same PC from my very first post on this site. The one and only thing that has ever gone wrong with that PC started with 'T' and ended with 'po'. In case you're wondering, that PC is still using the same $2 Thermal Master PSU I mentioned in that post!

It is about to reach 100,000h .
Seagate manufacture very good units.
Mine came to 100,000h, but I lived 3 years in a removable unit having power cuts every 5 seconds, I do not know how it got broken, I pass it to a computer with 73,000h and I get 100,000h until it started to have problems.

[eccerr0r]

Re: Hard drive reliability

I think I used the DOS utility to stop my WD Green from head unloading. Something like WDIDLE3.exe from WD's site, it should still be there I'd imagine.

---

I recall having a 7200.7 Seagate 40GB HDD at one point, it died after 25000 hours IIRC. Was accumulating tons of bad sectors, don't know exactly what went wrong with it.

[momaka]

Re: Hard drive reliability

I recall having a 7200.7 Seagate 40GB HDD at one point, it died after 25000 hours IIRC. Was accumulating tons of bad sectors, don't know exactly what went wrong with it.

Probably headstack went bad. I have a few dead 7200.7 in my junk box that I saved a while back when working at the IT dept of a medium size non-profit office. We had about 30-40 computers running those HDDs at the time, and over the period of a few years, these were the two that failed completely. So overall, they were pretty reliable HDDs.

[Pentium4]

Re: Hard drive reliability

Wow! Some high POH hard drives in here. I have found the 7200.7 to be pretty reliable, as well as WD Black. Hitachi drives also hold up well.

I have a 160GB 7200.7 that I think is finally on its last legs. It's the one I posted on the first page. It has 66,583 POH and it's starting to act funny. It has 5 reallocated sectors, 115 pending sectors, and the seek error rate number gets higher after every 10 seconds when transferring big files. I was just going to let it run in my room but it will make strange noises periodically so I think it's ready to retire. It had a long life, but not as long as kevins!

I have a Hitachi IDE drive in my Pentium III pfSense machine and it has 68,600 POH. It has 0 reallocated sectors and 1 pending sector, but after 6,000 hours that number hasn't increased. It also only has 116 power cycles, and the seek error rate is 0. I'm hoping it will last another 10,000 at least. Haha.

Here is a poor Raptor drive. It has a good life but I couldn't do anything to repair it. It spins up fine but when you try and run a test on it, it immediately fails. I tried plugging it into Windows and it won't show up in Disk Management. I was mostly bummed because I wanted to see the SMART statistics. The only thing I was able to see was the POH, which is a decently high number.


This drive when you plug it in won't make any noise but the logic board gets scalding hot really quickly. Couldn't figure out any stats on it.


This drive worked the first time I tested it, with around 23,000 hours if I remember correctly. It got stuck on a bad sector once and then the next time I tried to look at the data, it would spin up, then spin right down. Couldn't hear any arm movement.


These are the highest POH drives I've ever seen. They ran side by side in a Windows 2000 Advanced Server, well into 2015. They never failed! Although one of them has 8 reallocated sectors. Not too bad for that much use!


I have 80 drives I have tested so far and these are the only ones that have failed. I have a few more, but so far that's pretty good.

[goodpsusearch]

Re: Hard drive reliability

Deathstars lasting so long? So rare!

[Curious.George]

Re: Hard drive reliability

i have been testing drives this week for someone,
the newer ones are shit - no doubt about that.
if buying new, WD black, i dont know if they are good - but:
if they give a 5year warranty then they obviously put more attention into it's construction!

That's specious reasoning. They could also be total CRAP -- but, cost the manufacturer next to nothing to REPLACE! (the manufacturer places no value on your time or your data!)

if salvaging used drives then SCSI are best - i have over a dozen that are very old and still good.

One can argue that most SCSI/SCA/FC-AL drives were intended for server markets -- where the customer places a high value on availability and reliability. As such, they probably designed for (or, tested for) those qualities as a corporate user is far more likely to institute policies favoring (and AVOIDING) vendors and products based on their actual histories.

The "No-one-ever-got-fired-for-buying-IBM" argument.

Personally, I have drives in all interface technologies ranging from 40MB (that's an M, folks) to 4TB that I can spin up at will. So, I can't say any particular technology is "inherently bad" (or good!)

[I wouldn't be so quick to say that about "removable" technologies, though!]

[Pentium4]

Re: Hard drive reliability

I think they got the name "DeathStar" a little unfairly. I know they can't handle shock as much because of their glass(?) platters, but they prove to be dependable drives. If anyone here is interested, I am compiling a complete SMART statistics document of all the drives I have (100+), with their full model and manufactured date.

[Wester547]

Re: Hard drive reliability

Deathstars lasting so long? So rare!

The 180GXP and 120GXP series were generally much more reliable than the massively failure prone 75GXP and 60GXP series.

but the logic board gets scalding hot really quickly.

Without active cooling, especially in the duration of read/write operations, the microcontroller on a HDD PCB can dissipate quite a bit of heat very quickly (even more than the hermetic block!). But it still may be indicative of a fault with the drive.

I have a IC35L040AVVN07-0, firmware VA2OAG0A, which I've posted about before. It has a measly ~15,500 POH in comparison, but ~5,700 power on cycles with a min/max recorded lifetime temperature of 12C/54C in accordance to Victoria. It has 7 reallocated sectors and 0 pending, with 19 recorded errors in the S.M.A.R.T. log and 8 reallocation events. It passed a full disk scan without any errors. It was made in China on the 15th of October, 2002. I attached a brief recording of what it sounds like spinning up and down with its noisy and rather worn ball bearings (unlike the 180GXP series, the 120GXP series and earlier series use ball bearings rather than fluid dynamic bearings). The drive is noisier than the recording would have you believe.

BTW, the problem with the notorious Deskstars was not the glass platters but the antiferromagnetically coupled media. The pixie dust from the AFC media would sprinkle off the platters and onto the head stack, resulting in massive head crashes and a sea of bad sectors.

[momaka]

Re: Hard drive reliability

Here is a poor Raptor drive. It has a good life but I couldn't do anything to repair it. It spins up fine but when you try and run a test on it, it immediately fails. I tried plugging it into Windows and it won't show up in Disk Management.

Unfortunate.... but typical of old WD HDD failures - they work fine one day and then they don't. That's not to say they are/were less reliable than Seagate. Just their failure mode was more sudden. Most dead/dying Seagate HDDs I saw either were starting to make weird noises or racking up bad sectors and starting to crash/BSOD Windows, but still working overall. Same with Hitachi/IBM.

This drive when you plug it in won't make any noise but the logic board gets scalding hot really quickly.

That's an issue worth looking into. Might be able to revive the drive back if it's some faulty component on the board, like a shorted ceramic cap or similar.

These are the highest POH drives I've ever seen. They ran side by side in a Windows 2000 Advanced Server, well into 2015. They never failed! Although one of them has 8 reallocated sectors. Not too bad for that much use!
https://www.badcaps.net/forum/attach...1&d=1518209504

Over 100k hours? -NICE!!!
I also have a Hitachi 3.5" desktop HDD approaching close to 100k hours. See attachment IBM IC35L120AVVA07-0 (120 GB HDD Raduil PC).txt
At the time of taking that SMART info (about 2 years ago), the HDD had *only* 95.6 k power-on hours and 44 power cycles. When I bought it (used/refurbished drive from MicroCenter), I though the POH value was in minutes like my laptop's HITACHI DK23EA-20 - but NOPE. Those 95.6k POH are real.

I use that PC only for 2-3 weeks in the summer, though (with the PC being mostly off and stored away the rest of the time.) So that HDD is no longer going to rack hours like crazy. However, when I do use it then, I power-cycle it daily (sometimes even 2-3 times per day), so its power cycle count is going to increase much quicker now relative to the POH, just like with many of my other HDDs.

Yup, I am a hard "power cycler". Most of my PCs typically have almost as many power cycles as power-On hours.

Deathstars lasting so long? So rare!

I think they got the name "DeathStar" a little unfairly.

I agree with you, Pentium4.

Sure back in the day of good old reliable IDE drives, they were considered pretty bad. But in reality, even the worst "Death"stars of back then are no worse than your typical average drive today. Most lasted at least 10k POH. You'd actually be lucky to get that from a modern post-2015/2016 consumer mechanical HDD.

On that note, I have a 20 GB IBM Deskstar DTLA-307020 (75-GXP, I believe) that's still alive today. Granted I was very careful with it (to the point where I used to immediately jump out of my chair and power-down the PC - that was way way back in my early computer noobie days when I was young and had only ONE PC, so I cared very much for it and listened to every weird noise it made ... plus, 20 GB wasn't that little back then in the early 2000's). Here is its SMART log from around 2010 (which hasn't changed much, as the whole PC has been stored in my closet since 2008 or so, and used only briefly for a few times since then) - see IBM-DTLA-307020 (20 GB Duron PC).txt attachment.

Anyways, the thing about the 75-GXP line of IBM HDDs is they do NOT like heat. Once my drive heats up past 41C, it starts to do its click-beep head-seeking thing and eventually drops out and/or causes a BSOD until it is (hard) power-cycled. At that point, it may or may not a bad sector to its SMART count.

BTW, the problem with the notorious Deskstars was not the glass platters but the antiferromagnetically coupled media. The pixie dust from the AFC media would sprinkle off the platters and onto the head stack, resulting in massive head crashes and a sea of bad sectors.

Exactly.

And this happened the most when the drives heated up past a certain temperature, which caused several issues. Some of the biggest ones I hear were:
1) The different material expansion rate between the glass platters and the magnetic coating at high temperature caused the magnetic coating to become slightly "loose", and this is why the heads would often hit the platters.
2) To improve latency, IBM had heads flying closer to the platters. But when the heads sat idle over the same area for extended periods of time, it would cause a localized heating on the platters, thus running into issue 1) above again.
3) (and this may or may not be a myth), but IBM also had problems with the cache chips corrupting data - again, at least on the 75-GXP series.

I attached a brief recording of what it sounds like spinning up and down with its noisy and rather worn ball bearings (unlike the 180GXP series, the 120GXP series and earlier series use ball bearings rather than fluid dynamic bearings). The drive is noisier than the recording would have you believe.

Ah, now we are onto something here.
Let me see if I can make a sound recording of some of my HDDs. Anyone interested to hear what I have?