August 10, 200817 yr easiest way to post it is just the normal link like this: http://pastebin.com/md282b09 nothing fancier than that is required really. if you wanted to post the download as text file link it would be: http://pastebin.com/pastebin.php?dl=md282b09 As from drives a 3 are this: Model Number: SAMSUNG HD103UJ Firmware Revision: 1AA01110
August 11, 200817 yr hmmmm, my SAMSUNG HD103UJ's firmware revision is 1AA01112. I cannot make this drive spin down with any settings on the in-built way. Even not, if I using hdparm command and set the timer for 1 hour. If I run manually hdparm and set the timer around 20 minutes, than it is spinning time, but not stable. Sometime spins down, sometime not. In the meantime, the attached maxtor drive is always spinning down. I have tried in two different MB (Giga-byte GA-G33M-DS2R, ASUS P5B-E), same sympthoms. My friend of mine with an ASROCK MB and with an other HD103UJ is experiencing the same, as me, his drive is not spinning down, while his other drives is. So it means, that not my drive is faulty. Moreover the same happens with my SAMSUNG HD753LJ drive. I am really embittered. I am going to buy this product, as I like it very much and as it works like a charm in all other aspects, but I am not able to use it without spin down... I have a few more questions to you, NAS: - Are you sure, that your HD103UJ is spinning down? - Are you using the standard, in-built spin down method, or you are using individual spin down timers, playing with hdparm command? - What kind of MB you are using? - Are you using your drives in AHCI with native sata, or you are using legacy mode? - Which type and version of un-raid you are using? Thank you!
August 11, 200817 yr I have a few more questions to you, NAS: - Are you sure, that your HD103UJ is spinning down? - Are you using the standard, in-built spin down method, or you are using individual spin down timers, playing with hdparm command? - What kind of MB you are using? - Are you using your drives in AHCI with native sata, or you are using legacy mode? - Which type and version of un-raid you are using? Thank you! In order: Yes inbuilt Asus P5B as official board AHCI 4.3.3 Pro
August 11, 200817 yr Not too much differencies: SAMSUNG HD103UJ firmware version: yours 1AA01110, mine 1AA01112 Motherboard: yours P5B, mine P5B-E (in one of my setup) un-raid: yours 4.3.3 pro, mine 4.3.3 free I would say none of them should cause the different behaviour. I asked once, I ask again if under any circumtances, can it be caused by the free version? I know it shouldn't, and I am aware, that it is for sure not doing for purpose. But is it possible, that it has a bug in the code, which is limiting the capabilities of the pro version? I know it seems a stupid question, but based on my extensive testing you could follow above, and almost the same setup as NAS have, I think it might have a chance?!
August 11, 200817 yr Not too much differencies: SAMSUNG HD103UJ firmware version: yours 1AA01110, mine 1AA01112 Motherboard: yours P5B, mine P5B-E (in one of my setup) un-raid: yours 4.3.3 pro, mine 4.3.3 free I would say none of them should cause the different behaviour. I asked once, I ask again if under any circumtances, can it be caused by the free version? I know it shouldn't, and I am aware, that it is for sure not doing for purpose. But is it possible, that it has a bug in the code, which is limiting the capabilities of the pro version? I know it seems a stupid question, but based on my extensive testing you could follow above, and almost the same setup as NAS have, I think it might have a chance?! It should have nothing at all to do with the free version vs. paid version. Look in the syslog, if you see the hdparm commands setting the drive spindown, then they are being set. Joe L.
August 11, 200817 yr It should have nothing at all to do with the free version vs. paid version. Look in the syslog, if you see the hdparm commands setting the drive spindown, then they are being set. Joe L. Yes, in the syslog it is set, but I don't thought, that it is related to the seting of the spin down. My thought was, that maybe unRAID is staying the drives alive. But it was just a stupid guess. I have no more ideas, and it seems, that everybody else is out of ideas, as well.
August 11, 200817 yr Guys, it's getting more and more weird. My next finding is the following: Set spin down time to 20 minutes with hdparm (any more is not working at all) I include the command to go script. Booted up, after 20 minutes the drive goes down, as it should. Start to copy over something, the drive spin up well, the transfer is OK. After 30 minutes, it is still spinnig, so not spinned down. Than accessing the main page / or just simply running a smartctl -a -d ata /dev/sdX at consol. After 20 minutes it is goes down again, as it should. Start to copy over again something, the drive spin up well, the transfer is OK. After 30 minutes, it is still spinnig, so not spinned down. Than accessing the main page / or just simply running a smartctl -a -d ata /dev/sdX at consol. After 20 minutes it is goes down again, as it should. And this is 100% reproduceable! But remember, it is only for the Samsung disks, my maxtor always spinning down.
August 11, 200817 yr Guys, it's getting more and more weird. My next finding is the following: Set spin down time to 20 minutes with hdparm (any more is not working at all) I include the command to go script. Booted up, after 20 minutes the drive goes down, as it should. Start to copy over something, the drive spin up well, the transfer is OK. After 30 minutes, it is still spinnig, so not spinned down. Than accessing the main page / or just simply running a smartctl -a -d ata /dev/sdX at consol. After 20 minutes it is goes down again, as it should. Start to copy over again something, the drive spin up well, the transfer is OK. After 30 minutes, it is still spinnig, so not spinned down. Than accessing the main page / or just simply running a smartctl -a -d ata /dev/sdX at consol. After 20 minutes it is goes down again, as it should. And this is 100% reproduceable! But remember, it is only for the Samsung disks, my maxtor always spinning down. Than accessing the main page / Tell us more about how you are accessing "/" and what you are using to access it. Are you using "telnet" and typing "cd /" ? or just simply running a smartctl -a -d ata /dev/sdX at consol. invoking smartctl on a drive will spin up an otherwise idle drive. How are you determining the drive is spinning vs. sleeping?. Are you going by the "flashing indicator" on the unRAID web-page, or the light on the front of the disk? or listening to the disk spin? (The unRAID web-page does not refresh itself, so new status is only available if you refresh it) Also, if a program is accessing the disk from the lan, I don't think it will spin down. Disconnect the LAN cable, does it do the same thing? Joe L.
August 11, 200817 yr hdparm -C /dev/sdX will tell you the sleeping/not sleeping status of a drive without waking it up. (standby = sleeping) Joe L.
August 11, 200817 yr Hey Joe L., Thanks for the reply. Yes, I always check the status with "hdparm -C /dev/sdX". No, if I disconnecting the LAN cable, it doesn't make any difference. When I mentioned "Than accessing the main page", I understood accessing the Main tab on unRAID web interface. Of course I am aware that "smartctl -a -d ata /dev/sdX at consol" is waking up the drives. So what I am trying to describe is, that the inbuilt algorythm for spinning down is not working at all on my Samsung disk, but working on the maxtor at the same time. It is once an abnormal behaviour, but I want to find a workaround for that. This is where I get to manually set the spin down time to 20 minutes with hdparm -S240 /dev/sdx. It is working after bootup (hdparm is included in the go script) if I not touch the disk. After I copy over something to Samsung, it is not spinning down any more. But if I access the unRAID web interface (not doing anything, just access, and quit browswer), or run "smartctl -a -d ata /dev/sdX" at consol, it is again spinning down after 20 minutes. This is crazy!
August 12, 200817 yr Tested with 500GB SAMSUNG_HD501LJ as well. Working, spinning down with the inbuilt method, as it should. As I already mentioned, it is also spinning down with a 200GB Maxtor drive. Spinning down with old Maxtor IDE 160GB and Seagate IDE 200GB drives as well. So all in all, I think that the problem is related to the drives: I tested with 2 different Samsung 1000GB HD103UJ (both supplied with firmware: 1AA01112) I tried also with 2 different Samsung 750GB HD753LJ (both supplied with firmware: 1AA01107) None of the 4 drives was spinning down. Forum member: "NAS" reported that his Samsung 1000GB HD103UJ is working with firmware revision 1AA01110, so it seems Samsung screwed up something with the firmware in case of that type. Maybe HD753LJ is not working at all. As I spent the last two weeks with testing all the nights, so I am pretty confident that this is drive related. There are several drives working properly in my config, so it is not motherboard/bios/unRAID related. I tested with more drives from the same type. I think it can't be that all the four of the above drivers faulty. So, I would say be careful with these drive types.
August 12, 200817 yr hi everybody, this is my first post so I'll try to do it well (it is not easy for a French newb to be understood on a subject as technical and with such specialists) concerning the problem of spin down dd samsung, this is a problem fairly well known; I also have a NAS raidsonic 4220 B and I had the same problem with a dd samsung 500GB HD502IJ the problem was the same when i mounted it in unraid 4.33 but resolved through a command hdparm-S240 / dev / sd? that I have integrated into the GO file as Olympia said above we have to be careful when purchasing dd samsung (which are also very good) see here http://en.nas-4220.org/index.php?title=Harddrives
August 20, 200817 yr So, finally: I "trashed" the Samsung drive, and put in a WD10EACS 1TB Western Digital disk, and it is spinning down properly. All in all, I can confirm that Samsung HD103UJ and Sansung HD753LJ is not compatible with standard ATA commands, sent by hdparm. I tested them in an Ubuntu install as well, with the same results. I had more pieces of these disks, so it is not possible, that they had an individual fault. So I really don't understand how it is possible, that these type of disks are reported to work by a few forum member. However, some member says, that "maybe it is not spinning down at the time it should, based on the inactivity period set at settings tab, but later". I also experienced, that randomly they are spinning down in an absolutely unexpected moment (and in that case they stay down, so it is also not possible, that there are hidden activity, as in this case they should spun up again), but might only 6-7-8 or even 10 hours later, than they should. So I would recommend for those, who has these types of drives to check again, if they are spinning down properly just after 1 hour (in case it is set to 1). From my prospective, I would say, Samsung should stay in the vacum cleaner business!
August 20, 200817 yr All in all, I can confirm that Samsung HD103UJ and Sansung HD753LJ is not compatible with standard ATA commands, sent by hdparm. I can not agree with your findings in any way so far. I am not saying you are wrong but you have to be careful with such huge statements that could influence how many members spend their cash. I will be buying solely Samsung drives drives as they are cheap, no reliability problems so far and are noticeably cooler than WD in my setup. From my prospective, I would say, Samsung should stay in the vacum cleaner business! Lets balance that statement a bit by saying Samsung turnover over $1.3 billion a year they are not some nickel and dime company. To recap I am not trying to take away from what you are saying or say you are incorrect rather that YMMV and there are some serious pros to go with your con.
August 20, 200817 yr [To recap I am not trying to take away from what you are saying or say you are incorrect rather that YMMV and there are some serious pros to go with your con. I am afraid my english is not enough to fully understand this sentence. NAS, I honestly spent my evenings and weekends in the last 3 weeks testing out this. I close all the gap. Tell me how it is possible that my setup is working with all type of drives except these two (750GB and 1000GB Samsung). It is even working with an 500GB Samsung HDD. Can't be, that my config cause this, as in this case, same would be the case with the other disks. I tested with Ubuntu as well. And there are some forum members, experiencing very similar behaviour, just don't have enough time or energy to test it our, or they just don't care about spinning down. Again, I stated all this after 3 weeks long testing, not after one night. So tell me, where I am not careful.
August 20, 200817 yr I cannot tell you where your problem is i can only say that I cannot replicate this problem so it is not definitively a problem with the drives and may be a problem with something else completely differernt: disk9 ata-SAMSUNG_HD501LJ_S0MUJ1FP98---- * 488,386,552 24,706,544 2,046 23 0 disk10 ata-SAMSUNG_HD103UJ_S13PJDWQ32---- * 976,762,552 39,862,720 6,764 23 0 disk11 ata-SAMSUNG_HD103UJ_S13PJDWQ32---- * 976,762,552 229,314,892 33,852 897,283 0 disk12 ata-WDC_WD5000AAKS-00TMA0_WD-WCAPW196---- * 488,386,552 488,338,800 8 23 0 disk13 ata-SAMSUNG_HD103UJ_S13PJDWQ32---- * 976,762,552 20,784,144 502,405 60,903,344 0 disk14 ata-SAMSUNG_HD501LJ_S0Z4JDWPC0---- * 488,386,552 17,281,872 7,719 49,918,367 0 Or put another way: All in all, I can confirm that Samsung HD103UJ and Sansung HD753LJ is not compatible with standard ATA commands, sent by hdparm. Should be: I can confirm that Samsung HD103UJ and Sansung HD753LJ in my setup alone is not compatible with standard ATA commands, sent by hdparm. In other people setups it is completely fine.
August 20, 200817 yr He did not say you were wrong. He simply said that some other people, with other Samsung drives, connected to other hardware, have reported that their drives have spun down. It could be specific hardware version differences, even within a specific model of a drive... or firmware differences (even though the firmware seems to be identical) All he said was that on his hardware, using his disk controller, and his bios, and his motherboard, and his power supply... issuing a spin down timing command, that his SAMSUNG drives drives did spin down after the 1 hour idle interval. For that reason, and since the drives run cooler, and seem to be less of a load on the power supply, he will continue to use them and add more to his array. Clearly, you spent a lot of time experimenting. You found that 20 minute spin downs worked, but 1 hour and longer times did not. You know that 1 hour spin downs work with your other drives, and even an older Samsung. I'm glad you found a workable solution now that it is replaced. All your investigation effort may help other users of the same model Samsung drives. They will know the only solution for them is to RMA their drives (or use them where spin-down is not an issue) Hopefully you can RMA the drives that do not spin down and get replacements that do spin down when idle. Joe L.
August 20, 200817 yr Well said Joe. On pondering this the hhd daemon we are discussing in another thread may be a solution to this and other potential issues. Whilst timed spin downs may be problematic in some peoples setups ive never heard of anyone not being able to do a manual spindown with /usr/sbin/hdparm -y /dev/sd* >/dev/null. So once we have a daemon that essentially monitors usage we could be alot cleverer or at least allow users alot more scriptable control which in turn could resolve yet to be seen issues. In essence i dont think we should be trusting drives to spin down on their own i think we should be telling them to.
August 21, 200817 yr Well, okay, maybe I was too strict with Samsung, but I think you would think and state the same after 3 weeks struggling, than simply solve the whole issue by replacing the drive to a WD. And this issue should be exist for a long time (at least for my device versions), as the 750GB drive is about 3-4 month old, while the 1000GB was brand new. Joe L., some refining: I found 20 minutes idle interval working only in case I boot up in unRAID and do nothing. In this case, it was spinning down. After that, when I copy over something to that drive, it not spinned down again after 20 minutes, but maybe after 10 hours. If I set the idle interval for 1 hour, it is even not spinned down, just after reboot and completely inactive for 1 hour, but again, maybe spinned down after a few more long hours. Regarding RMA. To tell you the truth, I have not tried to RMA the drives. I cannot imagine the situation when I am standing there and say: the drive is working perfectly, spinning down in Windows (as it is, but I guess Windows is controlling the spin down interval on OS level, not on drive level), but not spinning down in Linux and Linux based unRAID, where spinning down delegated to the drive's firmware. I am pretty sure, that they would not replace the drive for this, and even if yes, I would not out of trouble, as I had 4 drives from them (purchased in different times in the last 3-4 month), with the same problem, so there is a big chance, that the next would have the same bug. Instead of that, I sold the 1TB drive on the used market. You can imagine that I am not too happy about the money spent on this replacement. NAS, yes, you are right, /usr/sbin/hdparm -y /dev/sd* was working all the time without any problem. In this case, if spinning time was controlled by the OS (unRAID), would defenitely would solve this issue, but it would be a workaround for the drives. Won't release Samsung in my eyes. But anyway, it would be a wonderful solution. Just one more question to you, are you 100% confident and can confirm, that your Samsungs are spinning down after the set inactivity interval and not later?
August 21, 200817 yr I have not sat an watched them so no i cannot say 100% certainty the spin down are 100% corext. However if they were wrong i would i have spotted them as I spend alot of time looking at unRAID and working on it.
August 21, 200817 yr NAS, if you are curious (I am), would you please do the following test? Reboot unRAID, copy over a 100MB file to all of your drive using /Disk* share instead of usershare, than start the stopper and see, if they are down after an hour.
August 21, 200817 yr I think time would be better spent trying to find a solution to your problem rather than me trying to prove i dont have one.
August 21, 200817 yr Well said Joe. On pondering this the hhd daemon we are discussing in another thread may be a solution to this and other potential issues. Whilst timed spin downs may be problematic in some peoples setups ive never heard of anyone not being able to do a manual spindown with /usr/sbin/hdparm -y /dev/sd* >/dev/null. So once we have a daemon that essentially monitors usage we could be alot cleverer or at least allow users alot more scriptable control which in turn could resolve yet to be seen issues. In essence i dont think we should be trusting drives to spin down on their own i think we should be telling them to. In the 2.4 kernel days, SATA support did not include the necessary API to implement all the 'hdparm' functionality; in particular, setting the spin down delays. Our solution then was a daemon that monitored i/o activity and explicitly spun drives down. We took this out sometime after switch to the 2.6 kernel. The reason to use the built-in standby timer in the drives is to handle the case: if the server software crashes for some reason, the hard drives would still spin down. Considering all the problems with spin-down/spin-up, I think we'll switch back to the daemon method.
August 21, 200817 yr Thats sounds very sensible. How about this though... do both. Set the built-in standby timer but set it long to say 4+ hours. Then we can use the control daemon for finer or more intelligent settings but with the fall back that should the daemon fail we know the drives will eventually spin down. If the daemon never crashes the built-in standby timers wont ever get used. One thing i would say is that per drive settings are a must. I have 1 or 2 drives that are hot hot hot and out of mazyness i spin down all drives when i see them cooking. Single drive settings would be a useful feature in this case and im sure a few other eventualities as well. Good to have you back Tom
August 22, 200817 yr Thats sounds very sensible. How about this though... do both. Set the built-in standby timer but set it long to say 4+ hours. Then we can use the control daemon for finer or more intelligent settings but with the fall back that should the daemon fail we know the drives will eventually spin down. If the daemon never crashes the built-in standby timers wont ever get used. I like this idea. One thing i would say is that per drive settings are a must. I have 1 or 2 drives that are hot hot hot and out of mazyness i spin down all drives when i see them cooking. Single drive settings would be a useful feature in this case and im sure a few other eventualities as well. You do have the option of setting shorter times in your go script. I actually have something in my go script to also initialize power up in standby. Note power up in standby via software control has only been verified (by me) with the WD green drives. Notice how I have the -S242 in there just in case ;-) I may change this to use serial numbers instead via /dev/disk/by-id so I do not have to worry about /dev/sdX position. This is my /boot/custom/etc/rc.d/S01-hdparm script #!/bin/bash for disk in /dev/sdc /dev/sdd /dev/sde /dev/sdf /dev/sdg do # S120 hdparm -S242 -s1 --yes-i-know-what-i-am-doing ${disk} hdparm -I ${disk} | egrep -i standby done
Archived
This topic is now archived and is closed to further replies.