loadme Posted January 26, 2011 Share Posted January 26, 2011 hello everyone. at the moment i use 4 drives parity: WD20EARS (2TB) disc1: WD20EARS (2TB) disc2: F3 (2TB) disc3: F4 (2TB) to expand my unRAID i took a look at 2TB drives and figured, that all those drives are still the cheapest ones and that i could do some small handmade benchmark (what worked best for me so far) to choose, what drive to rebuy. this is what i came up with: copy a 2GB file from windows over network to: sdb - parity sdd - disc1 - constant performance - 32 MB/sec sdc - disc2 - best performance, also constant - 37 MB/sec (the older F3 modell) sda - disc3 - worst performance - 0-38 MB/sec i continued this test 2 times with all drives and got the same results. disc3 (the newer spinpoint f4) seems to be buggy or badly configured or maybe not good in co-working with the WD20Ears parity disc while there are no write errors to this drive and the syslog seems fully friendly, it seems to perform very badly the drive starts the copy process with ~32 MB/sec but every 1-2 sec it drops down to 0-2 MB/sec for 2-3 seconds. later it goes back up to 38MB/sec and after 1-2 seconds it drops back down. (repeating this until the file is copied) is this common behaviour to the spinpoint f4? it could be unique to this specific disc and you have experienced the F4 as a perfect drive with unraid? right now i tend to buy one or two more F3 samsung for my expansion plans. would you consider replacing the f4, too? edit: hdparm: /dev/sda: (F4) Timing cached reads: 6564 MB in 2.00 seconds = 3287.67 MB/sec Timing buffered disk reads: 374 MB in 3.01 seconds = 124.36 MB/sec /dev/sdb: (parity WD20ears) Timing cached reads: 6212 MB in 2.00 seconds = 3110.52 MB/sec Timing buffered disk reads: 306 MB in 3.03 seconds = 100.89 MB/sec /dev/sdc: (F3) Timing cached reads: 6138 MB in 2.00 seconds = 3073.86 MB/sec Timing buffered disk reads: 322 MB in 3.01 seconds = 106.87 MB/sec /dev/sdd: (WD20ears) Timing cached reads: 6264 MB in 2.00 seconds = 3136.74 MB/sec Timing buffered disk reads: 254 MB in 3.01 seconds = 84.51 MB/sec /dev/sde: (unraid usb stick) Timing cached reads: 5570 MB in 2.00 seconds = 2789.06 MB/sec Timing buffered disk reads: 40 MB in 3.15 seconds = 12.69 MB/sec hdparm seems to read only. but the results are completly reversed. looks like the f4 has incredible reading results. Quote Link to comment
Joe L. Posted January 26, 2011 Share Posted January 26, 2011 What version of unRAID are these tests being made on? What hardware (motherboard, disk controller card?) How much RAM is installed? How "full" are the drives being written? (Outer cylinders on near empty drives will often have over twice the performance as inner cylinders) without knowing some of these, it is difficult to make any comparison. It sounds as if your disk3 may have some problems (or you are using an older version of unRAID). Are there any errors in the system log? Joe L. Quote Link to comment
loadme Posted January 26, 2011 Author Share Posted January 26, 2011 Hey Joe, thanks for the fast reply. System information: 2x1 GB DDR 2 667Mhz unRAID 4.5.6 unMenu and packages got updated today Mobo: n4l-vm-dh (2x SATA onboard, 1x SATA Raid onboard) CPU: Intel Core 2 Duo T7200 Box (Sockel M, 65nm, BX80537T7200) PCI (only PCI) controller with 1 SATA connector being used parity disc is connected directly to the board other discs are connected to the left 3 SATA slots. the raid sata slot is correctly recognized as a single disc. the F4 (problem-disk) is directly connected to the board. all disks are around 98% full. so yes, they maybe slower. but two perform as expected and the F4 seems to hang every now and than. syslog isnt showing anything at all during the copy process. the only orange/red msgs left when booting are: Jan 26 18:09:03 Tower kernel: ACPI Warning: Incorrect checksum in table [OEMB] - A9, should be A4 (20090903/tbutils-314) (Minor Issues) Jan 26 18:09:03 Tower kernel: pci 0000:00:1d.7: EHCI: BIOS handoff failed (BIOS bug?) 01010001 (Minor Issues) Jan 26 18:09:03 Tower kernel: ACPI Error (psparse-0537): Method parse/execution failed [\_PR_.CPU1._OSC] (Node f740e198), AE_ALREADY_EXISTS (Minor Issues) Jan 26 18:09:03 Tower kernel: ACPI: Marking method _OSC as Serialized because of AE_ALREADY_EXISTS error (Errors) Jan 26 18:09:03 Tower kernel: ACPI Error (psparse-0537): Method parse/execution failed [\_PR_.CPU1._PDC] (Node f740e180), AE_ALREADY_EXISTS (Minor Issues) Jan 26 18:09:03 Tower kernel: ACPI: Marking method _PDC as Serialized because of AE_ALREADY_EXISTS error (Errors) Jan 26 18:09:03 Tower kernel: processor LNXCPU:00: registered as cooling_device0 Jan 26 18:09:03 Tower kernel: ACPI Error (psparse-0537): Method parse/execution failed [\_PR_.CPU2._OSC] (Node f740e270), AE_ALREADY_EXISTS (Minor Issues) Jan 26 18:09:03 Tower kernel: ACPI: Marking method _OSC as Serialized because of AE_ALREADY_EXISTS error (Errors) Jan 26 18:09:03 Tower kernel: ACPI Error (psparse-0537): Method parse/execution failed [\_PR_.CPU2._PDC] (Node f740e258), AE_ALREADY_EXISTS (Minor Issues) Jan 26 18:09:03 Tower kernel: ACPI: Marking method _PDC as Serialized because of AE_ALREADY_EXISTS error (Errors) Jan 26 18:09:03 Tower kernel: ACPI: I/O resource w83627ehf [0x295-0x296] conflicts with ACPI region HWRE [0x290-0x299] (Minor Issues) Jan 26 18:09:03 Tower kernel: e1000e 0000:01:00.0: Warning: detected ASPM enabled in EEPROM (Minor Issues) Jan 26 18:09:07 Tower emhttp: shcmd (23): killall -HUP smbd (Minor Issues) Quote Link to comment
Rajahal Posted January 26, 2011 Share Posted January 26, 2011 Be warned that the Samsung F4s also have issues with data corruption. Upgrading the firmware is supposed to fix this, but there's no way to tell if the firmware upgrade was successful or not. Also, as Joe hinted at, if you have more than 2 GBs of RAM in your system, then a 2 GB test file isn't big enough, since your server's RAM can cache the whole transfer. Instead, try a 10 GB file or something that is guaranteed to be too large for your RAM. At least you are getting good performance on your disks 1 and 2. Quote Link to comment
loadme Posted January 27, 2011 Author Share Posted January 27, 2011 ok, so next disc should better be a spinpoint F3? damn. the WD20Ears is annoying with its jumper and my 20 tries to preclear it. the f4 has issues with corrupt data and seems to perform bad in my system and the f3 is doenst feels "new enough" Quote Link to comment
Rajahal Posted January 27, 2011 Share Posted January 27, 2011 With the new version of unRAID (4.7) the WD EARS no longer need a jumper. These drives remain my favorites. You can still use the Samsung F4s, but do so at your own risk. I personally will be staying away from these. F3s are actually great drives, nothing wrong them them. Just maybe a bit hard to find or a bit more expensive these days. WD EADS are the same story. Quote Link to comment
lionelhutz Posted January 27, 2011 Share Posted January 27, 2011 F3's are hard to find. You're looking at using a F4 or a EARS. You don't need the jumper anymore. I would hazard a guess that there is some kind of issue with that F4 drive. Others have had good luck and good speeds with them. Peter Quote Link to comment
loadme Posted January 28, 2011 Author Share Posted January 28, 2011 when installing unRAID 4.7 I simply remove the jumpers on EARS and everything remains the same? no need for preclear/formatting? really dont want to lose my data, because i "plugged the jumper" Quote Link to comment
Spectrum Posted January 28, 2011 Share Posted January 28, 2011 when installing unRAID 4.7 I simply remove the jumpers on EARS and everything remains the same? no need for preclear/formatting? really dont want to lose my data, because i "plugged the jumper" If your drive already has a jumper and in the array then leave the jumper alone. If you remove it the partitions will not be seen. If you are installing a NEW drive after upgrading to 4.7 then you can either preclear it with the new script and use the -A option, or add it to the array using the AF setting and unRAID will take care of it. Quote Link to comment
Joe L. Posted January 28, 2011 Share Posted January 28, 2011 when installing unRAID 4.7 I simply remove the jumpers on EARS and everything remains the same? no need for preclear/formatting? really dont want to lose my data, because i "plugged the jumper" If the drive is already assigned to the unRAID array, leave the jumper as it was before you upgraded. Changing it will lose all the data on that drive. Quote Link to comment
Rille_lkp Posted February 13, 2011 Share Posted February 13, 2011 Be warned that the Samsung F4s also have issues with data corruption. Upgrading the firmware is supposed to fix this, but there's no way to tell if the firmware upgrade was successful or not. Also, as Joe hinted at, if you have more than 2 GBs of RAM in your system, then a 2 GB test file isn't big enough, since your server's RAM can cache the whole transfer. Instead, try a 10 GB file or something that is guaranteed to be too large for your RAM. At least you are getting good performance on your disks 1 and 2. Samsungs released another firmware upgrade for these drives to resolve more issues and this update changes the firmware version unlike the first fix so that you can see that it has been updated. If the drive has been updated the firmware should be 1AQ10003. Update is available at http://www.samsung.com/global/business/hdd/faqView.do?b2b_bbs_msg_id=387 Quote Link to comment
dyrewolfe Posted February 13, 2011 Share Posted February 13, 2011 Samsungs released another firmware upgrade for these drives to resolve more issues and this update changes the firmware version unlike the first fix so that you can see that it has been updated. If the drive has been updated the firmware should be 1AQ10003. Update is available at http://www.samsung.com/global/business/hdd/faqView.do?b2b_bbs_msg_id=387 This update seems to apply to the /JP* variants. Does anyone know the difference between those and the other HD204UI variants (none, /Z4, /UZ4)? Quote Link to comment
Rille_lkp Posted February 13, 2011 Share Posted February 13, 2011 My drives doesn't have /JP in the name but just HD204UI. The update checks that you have the correct drive. I ran the update in a system with 9 Samsung drives where 4 of them are HD204UI and it only updates the correct drives. Says not supported for my other models. Quote Link to comment
jsdds Posted February 13, 2011 Share Posted February 13, 2011 Just to be clear then. Are you saying that the HD204UI drives, without the /JP, were in fact updated to the new firmware? Thanks. Quote Link to comment
Rille_lkp Posted February 13, 2011 Share Posted February 13, 2011 I can unfortunately not check the label on the disk itself without taking apart the server but SMART-information says "HD204UI" on model ID and the receipt says HD204UI, F4EG. There's no mention of /JP anywhere in the information. I also checked the firmwareversion before the update and they have been updated. When you run the update utility it checks all drives in your computer and updates compatible drives showing you if the update was successful or not. Quote Link to comment
jsdds Posted February 14, 2011 Share Posted February 14, 2011 That's great info! Thanks. Quote Link to comment
Rille_lkp Posted February 14, 2011 Share Posted February 14, 2011 No problem. I should also mention that i installed the first patch that didn't change the firmware version (http://www.samsung.com/global/business/hdd/faqView.do?b2b_bbs_msg_id=386) successfully on these drives when it was released. This patch is listed for the following models (note that there's no mention of /JP): F4EG HD204UI, HD204UI/Z4, HD204UI/UZ4, HD155UI, HD155UI/Z4, HD155UI/UZ4. Looks like Samsung have some wrong information regarding which drives the updates are for as i could install both patches but as their utility checks that you have the correct drive before it updates it should be safe to run the file in my understandning. EDIT: Better not use this update even though it works for me. Looks like the there are different updates for japanese drives and the update can't tell the models apart. I've emailed Samsung to check what they say. EDIT2: Samsung says JP firmware is fine for non JP-drives. Quote Link to comment
drf Posted February 14, 2011 Share Posted February 14, 2011 disc3 (the newer spinpoint f4) seems to be buggy or badly configured or maybe not good in co-working with the WD20Ears parity disc while there are no write errors to this drive and the syslog seems fully friendly, it seems to perform very badly the drive starts the copy process with ~32 MB/sec but every 1-2 sec it drops down to 0-2 MB/sec for 2-3 seconds. later it goes back up to 38MB/sec and after 1-2 seconds it drops back down. (repeating this until the file is copied) is this common behaviour to the spinpoint f4? Some other people also have those problems with F4 (have a look at the charts): http://hardforum.com/showpost.php?p=1036650393&postcount=1 http://hardforum.com/showpost.php?p=1036655921&postcount=14 Quote Link to comment
drf Posted February 15, 2011 Share Posted February 15, 2011 There have been tested 5 hard drives using different tests (mainly surface access times tests): 1. Samsung HD204UI in Verbatim 2 TB 3.0 USB enclosure. 2. Samsung SP1614N 3. WDC WD6400AAKS 4. WDC WD1600BEVS 5. Seagate ST320410A The tests results are in 149 files of the screenshots of these tests, available to download in this thread: http://forum.hddguru.com/surface-test-hdds-t18575.html 1. Generally Samsung drives (even the new 2TB Samsung HD204UI) seem to have considerably poorer quality of the surface comparing to the 3 years old (or so) Western Digital WDC WD6400AAKS. 2. I would like to see similar tests, in particular of other brands and models of 2 TB hard drives. Please upload your tests results in that thread (click the link above). 3. I wonder why in some hard drives slower access times can bee seen aligning in some diagonal patterns (in HD204UI) and why in some other drives (in WDC WD1600BEVS) a checkered pattern appears (only from the second half of the scan) which "shape" changes regularly depending on the part of the disk surface scanned, while in some other drives no such regular patterns are present? 4. Why is there such a difference in SMART status health notification between various software for Samsung SP1614N (it is red for most of the tests in HDD Health and everything OK in HD Tune)? There is also a significant difference in MHDD scan result and HDDScan result for SP1614N. 5. BTW, it took ca. 31 h to test Samsung HD204UI in Verbatim 2 TB 3.0 USB enclosure on USB 2.0 port. Quote Link to comment
Superorb Posted February 16, 2011 Share Posted February 16, 2011 I had read about the F4 drives being slow if they are not aligned. Mine is aligned (on XP system) and is very fast, faster than my 7200rpm drives. I believe the JP1 just refers to F4 drives shipped to the Japanese market, and according to a response by Samsung the new patch can be used on non-JP1 drives. I'm just worried that applying the patch on a drive with data could lose the data, and I don't have another drive to move the existing data to. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.