pras1011 Posted November 19, 2010 Share Posted November 19, 2010 I need some HDDs for my Unraid server. I may get the WD20EARS or Seagate Barracuda LP but I am uncertain about this. Can someone tell me which one is better or are there any other ones I should go for? Link to comment
ohlwiler Posted November 19, 2010 Share Posted November 19, 2010 Buy which is cheapest. I'd wait until Friday, we should see prices dip below $80. I'd also include the WD EADS into the mix. Some people advocate upgrading the firmware on the Seagate, although I don't think it is necessary, if you do the hassle may sway you. The WD EARS drive requires a jumper if you don't have some kicking around, that my influence you. Link to comment
pras1011 Posted November 19, 2010 Author Share Posted November 19, 2010 The Seagate and the WD are £70 in the UK. I have also noticed that there is a "park issue" with the WD and this makes them wear out quicker. Can anyone confirm this? Link to comment
BRiT Posted November 19, 2010 Share Posted November 19, 2010 I have not seen the LCC (Load Cycle Count) issue on recent WD drive models, only the older models, we're talking before the EADS which was before the EARS. Link to comment
mbryanr Posted November 19, 2010 Share Posted November 19, 2010 Here is WD's product bulletin on the issue. http://wdc.custhelp.com/cgi-bin/wdc.cfg/php/enduser/std_adp.php?p_faqid=5357&p_created=1266947046&p_sid=BXxBGP5k&p_accessibility=0&p_redirect=&p_srch=1&p_lva=&p_sp=cF9zcmNoPTEmcF9zb3J0X2J5PSZwX2dyaWRzb3J0PSZwX3Jvd19jbnQ9MTQsMTQmcF9wcm9kcz0yMjcsMjk0JnBfY2F0cz0mcF9wdj0yLjI5NCZwX2N2PSZwX3BhZ2U9MSZwX3NlYXJjaF90ZXh0PWxvYWQgY3ljbGUgY291bnQ!&p_li=&p_topview=1 Link to comment
pras1011 Posted November 19, 2010 Author Share Posted November 19, 2010 Here is WD's product bulletin on the issue. http://wdc.custhelp.com/cgi-bin/wdc.cfg/php/enduser/std_adp.php?p_faqid=5357&p_created=1266947046&p_sid=BXxBGP5k&p_accessibility=0&p_redirect=&p_srch=1&p_lva=&p_sp=cF9zcmNoPTEmcF9zb3J0X2J5PSZwX2dyaWRzb3J0PSZwX3Jvd19jbnQ9MTQsMTQmcF9wcm9kcz0yMjcsMjk0JnBfY2F0cz0mcF9wdj0yLjI5NCZwX2N2PSZwX3BhZ2U9MSZwX3NlYXJjaF90ZXh0PWxvYWQgY3ljbGUgY291bnQ!&p_li=&p_topview=1 I am not a techie, so is this bulletin good or bad news? Link to comment
BobPhoenix Posted November 19, 2010 Share Posted November 19, 2010 Here is my take on the article: Options: 1. Basically telling you how to keep the unRAID logs from showing the loads and unloads - I.E. ignore the issue. 2. How to use HDPARM to turn off the head parking (and other power saving features) to eliminate the load and unload cycles. This may have to be done on each boot of unRAID - don't know. 3. Use WD utility program to turn off the head parking (and other power saving features) directly on the drive. Shouldn't have to repeat this on each boot of unRAID - it's a one time thing. Just my 2¢. Link to comment
pras1011 Posted November 19, 2010 Author Share Posted November 19, 2010 Which should I get the Seagate or the WD? Link to comment
BRiT Posted November 19, 2010 Share Posted November 19, 2010 BobPhoenix, your take on Option 1 is wrong. What Option 1 from the Tech Bulletin does is remove unneeded writes to the drive, thus preventing it from spinning up. It does not IGNORE the issue, it prevents a lot of the causes of the issue. It also lets Linux keep the writes in a buffer until it decides to flush it, as opposed to forceably flushing after every single write. They're really steps that a Linux system should employ anyways regardless of drives being used. As it stands, Option 1a will not impact HDD's in standard unRAID systems as it uses a RAM-based filesystem and does not log to them. For what it's worth, here's the values from my three 1.3 years old EADS drives, all of which have less than 1 Load_Cycle_Count every 10 hours: 9 Power_On_Hours 0x0032 085 085 000 Old_age Always - 11631 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 65 193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 1143 9 Power_On_Hours 0x0032 085 085 000 Old_age Always - 11532 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 44 193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 866 9 Power_On_Hours 0x0032 085 085 000 Old_age Always - 11616 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 51 193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 899 Link to comment
BobPhoenix Posted November 19, 2010 Share Posted November 19, 2010 BobPhoenix, your take on Option 1 is wrong. What Option 1 from the Tech Bulletin does is remove unneeded writes to the drive, thus preventing it from spinning up. It does not IGNORE the issue, it prevents a lot of the causes of the issue. It also lets Linux keep the writes in a buffer until it decides to flush it, as opposed to forceably flushing after every single write. They're really steps that a Linux system should employ anyways regardless of drives being used. As it stands, Option 1a will not impact HDD's in standard unRAID systems as it uses a RAM-based filesystem and does not log to them. For what it's worth, here's the values from my three 1.3 years old EADS drives, all of which have less than 1 Load_Cycle_Count every 10 hours: 9 Power_On_Hours 0x0032 085 085 000 Old_age Always - 11631 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 65 193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 1143 9 Power_On_Hours 0x0032 085 085 000 Old_age Always - 11532 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 44 193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 866 9 Power_On_Hours 0x0032 085 085 000 Old_age Always - 11616 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 51 193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 899 Thanks. That was the one I was suspecting I might have wrong. Link to comment
BobPhoenix Posted November 19, 2010 Share Posted November 19, 2010 Which should I get the Seagate or the WD? I got the WD because I didn't want to deal with the hassle of the Firmware upgrade that MIGHT be necessary and I like WD better than Seagate. I had a number of 500GB or 1TB Seagate drives die on me 2-3 years ago and have had better luck with WD. Although I might change my mind with the number of WD drives that have died on me in the past 3 months. Link to comment
kaiguy Posted November 19, 2010 Share Posted November 19, 2010 My < 6 month old EARS drives have pretty high LCC counts. 9 Power_On_Hours 0x0032 097 097 000 Old_age Always - 2421 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 312 193 Load_Cycle_Count 0x0032 179 179 000 Old_age Always - 63653 9 Power_On_Hours 0x0032 097 097 000 Old_age Always - 2429 12 Power_Cycle_Count 0x0032 100 100 000 Old_age Always - 313 193 Load_Cycle_Count 0x0032 182 182 000 Old_age Always - 54493 Even my 2 week old EARS drives are higher than yours, BRiT. I'm thinking I need to use wdidle on mine... Link to comment
BRiT Posted November 19, 2010 Share Posted November 19, 2010 kaiguy, yes by all means use the w3idle program to get those LCC's under control. I'm surprised there's such a difference there. Link to comment
Spectrum Posted November 19, 2010 Share Posted November 19, 2010 This got me curious about my WD20EARS. for f in sda sdb sdg sde; do echo $f; smartctl -a -d ata /dev/$f|grep Load; done sda 193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 527 sdb 193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 513 sdg 193 Load_Cycle_Count 0x0032 199 199 000 Old_age Always - 3213 sde 193 Load_Cycle_Count 0x0032 200 200 000 Old_age Always - 404 Can you guess which one is the parity drive? These drive were all purchased at the same time and are < 6 weeks old. Makes me think I should run wdidle on /dev/sdg but leave the others alone. Thoughts? Link to comment
lionelhutz Posted November 20, 2010 Share Posted November 20, 2010 My WDC WD15EADS-00S2B0 has about 11000 LCC for 9500 hours. My others are well below that ratio. I'm not concerned about them. The post with 6 weeks of use is about 1000hrs. I would not even be too concerned about any of those numbers. Now kaiguy's numbers on the other hand do seem to be fairly high at about 20x the power on hours. When I was reading about this, I did find quite a few reports about drives that had reach millions of LCC's and they were still working. kaiguy's drives will likely reach about 200,000 per year which is way below what these people were reporting. I believe WD specs these drives for 300.000 LCC's which would be 1.5 years in kaiguy's case. However, I didn't find much that pointed to high LCC drives experiencing higher failure rates. So, I don't believe reaching 1000's or 10 of thousands isn't a big deal. If you're heading towards 100's of thousands then you should have some concern. I believe you actually set the head park delay longer, not disable it. Well, too long and it is effectively disabled I guess. Peter Link to comment
BobPhoenix Posted November 20, 2010 Share Posted November 20, 2010 Most of my EADS and EARS are between 2.5 & 4 times the power on except for this WD RMA'd EADS: Power_On_Hours 0x0032 097 097 000 Old_age Always - 2529 Load_Cycle_Count 0x0032 188 188 000 Old_age Always - 37168 Link to comment
f00kie Posted November 20, 2010 Share Posted November 20, 2010 So what's the verdict on the WD20EARS? I'll be building a 4-8 HDD system soon; I've been burned on Seagate's before, so are the WD Green Drivers just going to cause more problems? Should I be looking into Samsung? Link to comment
pras1011 Posted November 20, 2010 Author Share Posted November 20, 2010 Apparently the Samsung F4 are no good too!! Link to comment
JackBauer Posted November 20, 2010 Share Posted November 20, 2010 So what's the verdict on the WD20EARS? I'll be building a 4-8 HDD system soon; I've been burned on Seagate's before, so are the WD Green Drivers just going to cause more problems? Should I be looking into Samsung? I have sworn off the EARS (64MB cache model, right?)... The horrible reviews at newegg, then after gambling and getting a bad one... Really screwed me over. Link to comment
Giraffeninja Posted November 20, 2010 Share Posted November 20, 2010 My vote is the samsung f3's. 2 TB non advanced format, low power and runs cool. Link to comment
lionelhutz Posted November 20, 2010 Share Posted November 20, 2010 I have sworn off the EARS (64MB cache model, right?)... The horrible reviews at newegg, then after gambling and getting a bad one... Really screwed me over. Thd 2T options at the lowest price point are the WD EARS, Seagate LP or Samsung S4. The EARS reviews are a little better than the Seagate reviews and the Samsung F4 drives don't work right. So, either take your chances with the WD or Seagate or up your purchase price for a different model. Just know that the higher priced WD or Seagate models don't show much better reviews. I just noticed I left out a 0 on my LCC number. I still consider 11,000 LCC's fine though. Peter Link to comment
lionelhutz Posted November 21, 2010 Share Posted November 21, 2010 I actually should not say the F4's don't work right, you just can't align the file system to the 4k sectors. It won't make much difference for media files but could have an impact on reading and writing lots of small files. Peter Link to comment
pras1011 Posted November 21, 2010 Author Share Posted November 21, 2010 I think I will get the Samsung F3! Link to comment
Giraffeninja Posted November 21, 2010 Share Posted November 21, 2010 I think that's a great choice! Do it while you can, like the tiger it is beginning extinct from all these poachers. Link to comment
MortenSchmidt Posted November 21, 2010 Share Posted November 21, 2010 I've got a few green drives. One which is in the array, is used as a DL disk for SABnzbd and Transmission (and seeding in Transmission). It's got 5648 hours and LCC of 89.567. The other drives used to store movies and backups have next to none. One example is an identical EADS 2TB drive with 10323 hours and LCC of 1266. Both disks have only ever been used in unRAID. So, how much LCC is accumulated seems to be dependent on the use for that particular disk in the array. kaiguy, yes by all means use the w3idle program to get those LCC's under control. I'm surprised there's such a difference there. BRiT, How did you say one could use that w3idle thing? A forum search came up blank - I'd appreciate a DL link and example line for the GO script. Thanks! UPDATE - I guess you mean wdidle. Is there a way to run that headless or will I need to drag a monitor down to the basement ? Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.