Smitty2k1 Posted June 25, 2018 Share Posted June 25, 2018 Howdy folks, I just upgraded an old 2TB drive to a new 8TB drive out of a WD EasyStore. The drive is model number WD80EMAZ, which is the one that has 3.3v pin issues. It turns out my PSU, a Silverstone SX500-LG isn't compatible with these drives, so I've used a Molex to SATA power adapter and got the drive working fine (half of the drives in my case are attached via a Molex to SATA strand from Silverstone).  Anyways, the drive precleared with no errors and I successfully rebuilt the array after upgrading the 2TB to 8TB drive. Read and write speeds are just fine, however the drive WILL NOT SPIN DOWN. It is set to default spindown like all my other array drives, but it never spins down. Manual spindown does seem to work. Is it possible the inability to spindown is related to the 3.3v issue many people have with these drives?  Logs don't show anything out of the ordinary. Link to comment
Frank1940 Posted June 26, 2018 Share Posted June 26, 2018 Relax for about a day. It could be that the preclear signature was incorrect or incorrect and it is simply updating drive so that parity is correct.  Link to comment
PSYCHOPATHiO Posted June 26, 2018 Share Posted June 26, 2018 just a heads up, be careful about the molex cable. they are really not so relaiable and could melt under high loads, just try getting a different compatable power supply. Link to comment
pwm Posted June 26, 2018 Share Posted June 26, 2018 4 hours ago, PSYCHOPATHiO said: just a heads up, be careful about the molex cable. they are really not so relaiable and could melt under high loads, just try getting a different compatable power supply. Â It's just one type of Molex to SATA connector type that seems to be problematic. A good Molex to SATA is no different from any other xx to SATA - in the end, you always needs a cable that ends with a SATA power plug. Link to comment
Frank1940 Posted June 26, 2018 Share Posted June 26, 2018 6 hours ago, PSYCHOPATHiO said: just a heads up, be careful about the molex cable. they are really not so relaiable and could melt under high loads, just try getting a different compatable power supply.  I assume that you are talking about this issue:       https://www.youtube.com/watch?v=TataDaUNEFc  It is real but to my knowledge, it appears to be limited to this type of design/construction.  Link to comment
PSYCHOPATHiO Posted June 26, 2018 Share Posted June 26, 2018 1 hour ago, Frank1940 said: Â Â Â Â Â https://www.youtube.com/watch?v=TataDaUNEFc yup, i love his videos. He knows what he talks about. Link to comment
pwm Posted June 26, 2018 Share Posted June 26, 2018 I wonder if the issue has been corrected - I have seen PSU with SATA connectors of both types but not seen issues with fire. So SATA power connectors of both types should work.  Maybe some factory just sent out a million bad cables that have now been sold out or removed from the market. No way for us to know.  Without a way to figure out what is a good or bad plug, we just have to settle for the type of plug that we can identify from the outside.  My guess is that Molex defaults to slightly thicker wires, and that the faulty adapter cables were produced with a SATA power connector that could not correctly handle the wire thickness. Lots of Molex-to-SATA are probably produced based on the same material used when making Molex-to-Molex - but with the device connector replaced with a SATA plug. Link to comment
Fireball3 Posted June 26, 2018 Share Posted June 26, 2018 14 hours ago, Smitty2k1 said: It is set to default spindown like all my other array drives, but it never spins down. Do you have a log of the SMART attributes of that drive? Link to comment
Frank1940 Posted June 26, 2018 Share Posted June 26, 2018 41 minutes ago, pwm said: My guess is that Molex defaults to slightly thicker wires, and that the faulty adapter cables were produced with a SATA power connector that could not correctly handle the wire thickness. Lots of Molex-to-SATA are probably produced based on the same material used when making Molex-to-Molex - but with the device connector replaced with a SATA plug.   From some other posts about this problem, it was not a wire gauge issue. The theory was that the molding encapsulation process was 'pushing' the wires closer together but not quite touching. Subsequent wire movement from normal handling and installation moved them even closer. Eventually, some small percentage of connectors would have a wire strand touch and create a temporary short. (It is all down hill from there. The plastic used will carbonize when subjected to high temperatures and that forms a low resistance current path. Since it is not a true short circuit with currents beyond the PS rating, the PS remains operational. The heat generated will cause the plastic to burn. )  Link to comment
JorgeB Posted June 26, 2018 Share Posted June 26, 2018 I've used hundreds of molex to SATA adapters for years, both at home and at work, and never had any issues with fire, obviously it might happen with a few, but most are OK, it's like cars, once in awhile a car catches fire, doesn't mean it will happen to all of them. Link to comment
pwm Posted June 26, 2018 Share Posted June 26, 2018 41 minutes ago, Frank1940 said:   From some other posts about this problem, it was not a wire gauge issue. The theory was that the molding encapsulation process was 'pushing' the wires closer together but not quite touching. Subsequent wire movement from normal handling and installation moved them even closer. Eventually, some small percentage of connectors would have a wire strand touch and create a temporary short. (It is all down hill from there. The plastic used will carbonize when subjected to high temperatures and that forms a low resistance current path. Since it is not a true short circuit with currents beyond the PS rating, the PS remains operational. The heat generated will cause the plastic to burn. )   Either the SATA power connector was faulty or used with wrong cables.  Because the SATA power connector can't know what is fitted at the other end of the cable. The only possible other factor here is if they incorrectly aligned the cables caused by the Molex connector not containing any 3V3 orange wires.  Anyway - it's a production quality issue relating to some specific production of adapter cables. Hopefully corrected since a number of years.  In the general case, there isn't a problem having Molex to SATA adapter cables. I have had a large number of them without issues and so have millions of other uses. But it's enough that a batch of 10k or 100k of faulty cables hits the market for Youtube to be filled with videos of the resulting failures. Link to comment
Frank1940 Posted June 26, 2018 Share Posted June 26, 2018 If you watch the video, you will find that the problem only occurred with on one type of connector design-- The one with the wires molded into the plastic. The snap together connector has not caused any problems. So, yes,even the vast majority of Molded SATA ends have not had a problem but some percentage of them have had issues where the plastic has smoked and possibly had a open flame. So the best advice would not be to completely avoid ALL Molex-to-SATA connectors but to avoid the ones that used the molded connectors. But how do you identify which ones might have an issue... At the cost of these connectors, I would recommend removing any of the molded ones that you might have in use and replace with the snap-assembled ones. I know that I did!  Remember that even if the connector does not burst into flame, the fire retardant in the plastic probably produces highly toxic flumes! Link to comment
pwm Posted June 26, 2018 Share Posted June 26, 2018 2 hours ago, Frank1940 said: If you watch the video, you will find that the problem only occurred with on one type of connector design-- The one with the wires molded into the plastic.  It's the old problem that correlation does not imply causation. There must be lots of connectors of that design that does not have issues. But since we don't know how to separate different brands or production runs we can't identify which connectors that comes from the faulty production run and which connectors just happens to look the same without having issues with internal short-circuit. Link to comment
Smitty2k1 Posted June 26, 2018 Author Share Posted June 26, 2018 I'm using a Molex to SATA cable like this one: https://www.silverstonetek.com/product.php?pid=384&area=en  It's from the PSU manufacturer, I'm not concerned about the fire risk that the cheap ones had.  SMART attributes attached   WDC_WD80EMAZ-00WJTA0_7JKPTLPC-20180626-1541.txt Link to comment
JorgeB Posted June 27, 2018 Share Posted June 27, 2018 SMART looks fine, I remember someone posting recently that 10TB WD Reds don't spin down, not sure if some 8TB models could be affected. Link to comment
Fireball3 Posted June 27, 2018 Share Posted June 27, 2018 What ist the setting of Auto Offline Data Collection enabled or disabled?  Edit: Sorry, my fault, missed @Smitty2k1 post.  Set it disabled and see if it helps. smartctl -o off /dev/your_drive  Link to comment
Zonediver Posted June 27, 2018 Share Posted June 27, 2018 3 hours ago, johnnie.black said: SMART looks fine, I remember someone posting recently that 10TB WD Reds don't spin down, not sure if some 8TB models could be affected. Â The 10TB WD-Red support APM with hdparm -B /dev/sdx you can set the APM-Level. Standart Value should be 164. As far as i know, this value must be between 1-127 or 255 (deactivated APM) to activate the spindown. I dont have a 10TB Red at the moment so i can't test this. Maybe someone can test it and give us a feedback. Link to comment
Fireball3 Posted June 27, 2018 Share Posted June 27, 2018 His SMART report says: APM level is: 164 (intermediate level without standby) Â It can also be set with smartctl -s apm,255 Some testing is required: Quote If a value between 1 and 254 is provided, it will attempt to enable APM and set the specified value, 'off' disables APM. Note the actual behavior depends on the drive, for example some drives disable APM if their value is set above 128. Values below 128 are supposed to allow drive spindown, values 128 and above adjust only head-parking frequency, although the actual behavior defined is also vendor-specific. Â Link to comment
Zonediver Posted June 27, 2018 Share Posted June 27, 2018 5 minutes ago, Fireball3 said: It can also be set with smartctl -s apm,255 Â Â Is this persistent? Link to comment
Fireball3 Posted June 27, 2018 Share Posted June 27, 2018 14 minutes ago, Zonediver said: Is this persistent? I don't know. Â Edit: Some additional info on APM levels. https://www.freeminded.org/index.php/2012/04/hdparm-power-down-drives-when-not-in-use/ Link to comment
Fireball3 Posted June 27, 2018 Share Posted June 27, 2018 I wonder what unRAID is doing when we set the "drive spindown"!? Is is also using the drive's apm settings or is it handled different? Link to comment
Zonediver Posted June 27, 2018 Share Posted June 27, 2018 15 minutes ago, Fireball3 said: I wonder what unRAID is doing when we set the "drive spindown"!? Is is also using the drive's apm settings or is it handled different? Â Good question... all my Reds (3 and 4 TB) doesn't support APM... so i can't test it. Link to comment
Smitty2k1 Posted June 27, 2018 Author Share Posted June 27, 2018 Hey everyone, I think I've found something. The drive successfully spins down, but it seems to spin UP at inappropriate times, which is why I thought it wasn't spinning down.    For example, I just downloaded a file to my server which should use the cache drive. It does in fact use the cache drive but the data drive in question is also spinning up. Also, browsing files (with the cache files plugin) this data drive spins up.  The 'Dashboard' tab of the unRaid GUI shows no streams to any of my shares, and the read/write count of this drive does not go up, but it seems to spin up whenever anything else on the server is accessed. Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.