|
HalloKitty posted:They just need to stop selling the drives full stop. Call them Greens, and scrap the entire Red lineup, and just move people to "Red Pro". I’m hoping crazy n-layer NAND will soon become cheap enough to use it on a home NAS level. SMR drives would probably work alright as cold backups of that. CMR won’t come back to consumer drives but I’m sure they’ll be happy to sell you prosumer/”creator” drives for twice the price because
|
# ? Apr 21, 2020 21:22 |
|
|
# ? Jun 5, 2024 19:34 |
|
If anything, this whole SMR thing feels like such a weird self own on WD's part. The people buying shuckables were getting more for the same price, and getting CMR, not SMR. The people who were being smart and buying their cheaper product ended up with a more quality product than the ones buying the real name brand, and now you're just going to push people towards the shuckables.
|
# ? Apr 21, 2020 22:15 |
|
https://arstechnica.com/information-technology/2020/04/seagate-says-network-attached-storage-and-smr-dont-mix/ Seagate at least promises that their ironwolf drives will stay CMR.
|
# ? Apr 21, 2020 22:34 |
|
Ironwolves are the only drive I've ever had 100% failure rates with but that alone would be enough for me to try them again assuming cheaper sources dry up.
|
# ? Apr 21, 2020 23:04 |
|
IOwnCalculus posted:Ironwolves are the only drive I've ever had 100% failure rates with but that alone would be enough for me to try them again assuming cheaper sources dry up. I looked at the most recent backblaze data and whatever Seagate drives they use only have a slightly higher failure rate vs HGST and Toshiba. But I don't know if those are ironwolfs or not E: ooh, even better is that their datasheets explicitly call out CMR: https://www.seagate.com/internal-hard-drives/hdd/ironwolf/ Every single ironwolf from 16 TB down to 1 has it Yaoi Gagarin fucked around with this message at 23:20 on Apr 21, 2020 |
# ? Apr 21, 2020 23:15 |
|
VostokProgram posted:I looked at the most recent backblaze data and whatever Seagate drives they use only have a slightly higher failure rate vs HGST and Toshiba. But I don't know if those are ironwolfs or not
|
# ? Apr 21, 2020 23:57 |
|
DrDork posted:The HP workstation series (Z440, etc) are also decent alternate options. They tend to sell for a bit less than the Dell or Lenovo options, probably due to being less popular on some of the DIY build sites, but they work just fine. Biggest downside to them is they tend not to have IPMI, so you'll need to manually set things up at least once. They also use a custom power cable like some of the Dell's, so same deal there ($20-$40 adapter) if you wanted to move the guts into your own case, but I've found mine to be reasonably quiet as-is. sharkytm posted:They also use really odd mounting screws with built-in shock absorbers, some of which are nearly impossible to find and are often discarded by resellers. You can buy them online or 3D print your own, but it's a word of warning to potential buyers. I've got several HP workstations, and I've had to 3D print spacers for the drive cages to work properly. There are even 2 different ones for 2.5" drives, depending on the model of computer... I'm happy with the computers, but the screws are a pain in the rear end. My SSD in the Elitedesk 800 G2 SFF are just double-sided taped in place because I couldn't be bothered to dimension the second type of screw spacer. I've got an EliteDesk 800 G2 tower too, and it takes the same dumb system. I find the key is to find one that's more or less matx sized; the lower-end precisions use straight up matx boards, with their custom power cable, but it fits in a regular sized case just fine, uses a standard io plate, etc. At least in this application, if I was going dual cpu (eg a lenovo p500) it would be a different situation... 5 in 3 would be fine for drives if you can live with 5 drives.
|
# ? Apr 22, 2020 00:40 |
|
VostokProgram posted:I looked at the most recent backblaze data and whatever Seagate drives they use only have a slightly higher failure rate vs HGST and Toshiba. But I don't know if those are ironwolfs or not I bought a bunch of Ironwolves in the 4-6TB range - I can't remember anymore but I'm sure it's in my post history here - and literally every single one of them would start making GBS threads up syslog with errors during a nwipe run, on a timeframe ranging from minutes to hours, combined with slowing to an absolute crawl. It was bizarre, I have no explanation for it, but I just returned them all and went with Toshibas instead.
|
# ? Apr 22, 2020 01:50 |
|
So in light of the WD Red debacle, is it reasonable to assume purples would remain CMR (albeit with a performance hit on reads due to the different firmware?)
|
# ? Apr 22, 2020 05:33 |
You know, I would like to see a WORM-like test of a zfs pool of PMR drives in raidz3 filled to 80% and then scrubbed and punished under some heavy write loads, and scrubbed again. If I understand the code right, it's the "one disk is SMR while the others are PMR" scenario that results in the drive being faulted. I could live with cheap SMR drives for my offline backup array, but I don't have the budget to do this kind of a test.
|
|
# ? Apr 22, 2020 09:38 |
|
I have two 10TB Ironwolves that work (although they needed a firmware update that was not exactly well publicised). I still prefer the white label shucked hgst engineered drives overall, but it's nice Seagate have taken a stand (for now) HalloKitty fucked around with this message at 11:19 on Apr 22, 2020 |
# ? Apr 22, 2020 10:47 |
|
Anybody else have shucked 14TBs yet? I'm wondering if these are SMR, because my 2 14TB drives have consistently 3-4x Utilization% reported by my Synology. This is with no data transfer currently happening except for seeding some torrents (at only 500KB/s). edit: Am I an idiot, and it's just because those 2 14TB drives are probably being used as the 2 parity drives? Brain Issues fucked around with this message at 12:42 on Apr 22, 2020 |
# ? Apr 22, 2020 12:32 |
|
Brain Issues posted:Anybody else have shucked 14TBs yet? In the Synology units parity is striped across all drives, there are no dedicated parity drives as there are in systems such as unraid or snapraid or a raid4 system. There may be another reason why the drives are showing activity. How long have the drives been in the system? Synology uses mdadm and btrfs(optionally) to build arrays, so if the drives are showing activity it may still be building the array. You could try using the console and checking /proc/mdstat (cat /proc/mdstat) and seeing if any of the arrays are still being built.
|
# ? Apr 22, 2020 13:31 |
|
Nulldevice posted:In the Synology units parity is striped across all drives, there are no dedicated parity drives as there are in systems such as unraid or snapraid or a raid4 system. There may be another reason why the drives are showing activity. How long have the drives been in the system? Synology uses mdadm and btrfs(optionally) to build arrays, so if the drives are showing activity it may still be building the array. You could try using the console and checking /proc/mdstat (cat /proc/mdstat) and seeing if any of the arrays are still being built. I put the two 14tb drives in about 2 months ago, and converted from SHR-1 to SHR-2. The conversion took 3 weeks to finish building the array. It is no longer building the array. code:
Do you have some documentation supporting that Synology doesn't use dedicated parity drives for SHR-1/SHR-2? Because it doesn't seem like that is what Synology is doing to me. Brain Issues fucked around with this message at 13:43 on Apr 22, 2020 |
# ? Apr 22, 2020 13:35 |
|
Brain Issues posted:I put the two 14tb drives in about 2 months ago, and converted from SHR-1 to SHR-2. The conversion took 3 weeks to finish building the array. It's just the underlying technology. You're seeing it right there in the output of the mdstat. Each array is showing either raid6 or raid1, not raid4 (dedicated parity). Synology doesn't give you the option to use a dedicated parity drive. Which disks are serving up/downloading the torrents?
|
# ? Apr 22, 2020 13:51 |
|
Nulldevice posted:Which disks are serving up/downloading the torrents? How can I tell? They're all part of 1 volume.
|
# ? Apr 22, 2020 13:57 |
|
Brain Issues posted:How can I tell? They're all part of 1 volume. Hmmm, iostat if it's available in the shell could tell you which array is servicing the program for your torrents. Unfortunately my Synology is at my folks place 550 miles away so I can't experiment. I haven't heard anything about the 14TB drives being SMR at this time, so I wouldn't worry about it just yet. I know my 1019+ with 5x12TB is always doing something (can hear the disk activity) but it's low level stuff, similar or lower to your 14TB drives. I just don't worry about it. When I retrieve the unit I'm going to see what the activity is.
|
# ? Apr 22, 2020 14:39 |
|
I shucked four 14TB drives the other day and did a Windows full format. Task manager was showing all four plugging along at 145MB/s, which multiplied by 60*60*24 = ~12.5TB... so maybe? I don't know if writing all 0s doesn't cause the cache thing to spaz out, but as someone who doesn't know what they're talking about, it passes the sniff test for me. Everything I've seen online has said that the Easystores are all CMR.
|
# ? Apr 22, 2020 15:21 |
|
I thought the big issue with SMR is that the reallocation / garbage collection activity is "silent" to the system / user, where the drive does its poo poo but to everything else it looks hung?
|
# ? Apr 22, 2020 15:23 |
|
Oh great a drive died guess I'll come in here to see what to...get... poo poo hell of a time for an 8tb seagate ironwolf that's only 1 year old to get 8 unreadable sectors. My hdd luck is real bad (though probably some of this is drive temperature inside the dell t20 being godawful and not fixable even after I strapped a second fan to it)
|
# ? Apr 22, 2020 16:59 |
|
Can any of you help me figure out my DNS problem with Unraid without putting a manual host entry on every device I try to manage it from? Right now I try to navigate to the web UI at 192.168.1.5*, but it immediately forwards me to tower.local which refuses to resolve. Right now my router DHCP hands out the lancache docker as primary DNS and the router itself as secondary DNS. *not the actual IP I use.
|
# ? Apr 22, 2020 17:23 |
|
CopperHound posted:Can any of you help me figure out my DNS problem with Unraid without putting a manual host entry on every device I try to manage it from? step 1: doing DNS that does two separate things is unlikely to work properly. (you'll be likely to get things that don't hit the cache because DNS primary/secondary doesn't work the way you think it does) my guess is your lancache docker is being dumb and is aggressively trying to resolve things (including *.local) and doesn't know about unraid
|
# ? Apr 22, 2020 22:42 |
|
California goons, I have an enterprise 6TB SAS (SATA compatible?) disk for sale over here for cheap, I found it in my closet https://forums.somethingawful.com/showthread.php?threadid=3920975
|
# ? Apr 23, 2020 01:07 |
|
FYI SAS/SATA compatibility only works one way - SAS controllers can run either SAS or SATA drives. SAS drives only work on SAS controllers. Which a lot of people in this thread have.
|
# ? Apr 23, 2020 01:15 |
I forget, is SAS required when you're using a disk chassis with 2 SAS expanders connected to two separate SAS HBAs via SFF-8088 (ie. A+B failover)? The disk-shelves I've gotten have apparently got two SAS expanders which I believe are separately connected to all ports internally.
|
|
# ? Apr 23, 2020 14:21 |
|
SAS is required depending upon the controllers built-in to the backplane but nearly all except certain NAS and SAN manufacturers support SATA last I saw given it costs more to block SATA drives given SATA is a subset of SAS standards. I think Isilon and EMC shelves block bog standard SATA drives in their backplanes.
|
# ? Apr 23, 2020 14:53 |
|
D. Ebdrup posted:I forget, is SAS required when you're using a disk chassis with 2 SAS expanders connected to two separate SAS HBAs via SFF-8088 (ie. A+B failover)? SAS controllers, yes. They'll probably be just fine with SATA disks though. I have a mix of both in my DA4243 shelf. If the SAS expanders are set up for redundancy, you might only be able to use one instead of both since multipathing is something only SAS drives can do.
|
# ? Apr 23, 2020 15:12 |
|
I'm going to get back into having a NAS after my homemade one died last year. I'll mainly use the NAS as a Plex server, and I'll be sharing my library with about a dozen friends. This time, I'd like to buy a consumer grade NAS, and I'm looking at the QNAP TVS-682-i3-8G-US 6 Bay, it's going for $1,300 on Amazon. Is this a good buy? Edit: I went ahead and ordered this guy: https://www.newegg.com/qnap-tvs-682-i3-8g-us/p/N82E16822107339?Item=N82E16822107339 It seems to be the same model as on Amazon but with a newer i3-7100 CPU, awesome. I also ordered 3 Seagate IronWolf 10TB NAS drives to go along with it, and I have a spare 970 Evo 500 GB to use as a cache drive. Nfcknblvbl fucked around with this message at 19:34 on Apr 23, 2020 |
# ? Apr 23, 2020 16:29 |
IOwnCalculus posted:SAS controllers, yes. They'll probably be just fine with SATA disks though. I have a mix of both in my DA4243 shelf. I think the multipath GEOM module in FreeBSD makes it possible to use any disks that are multipathed irrespective of their interface; it certainly works for SCSI, SAS, and iSCSI as well as FiberChannel - so I don't see why it shouldn't work on SATA either as long as the controllers presents the disks. I will try to remember to report back once I get hardware to play with, if it's of interest. BlankSystemDaemon fucked around with this message at 18:42 on Apr 23, 2020 |
|
# ? Apr 23, 2020 18:32 |
|
Buff Hardback posted:step 1: doing DNS that does two separate things is unlikely to work properly. (you'll be likely to get things that don't hit the cache because DNS primary/secondary doesn't work the way you think it does)
|
# ? Apr 23, 2020 19:00 |
|
D. Ebdrup posted:Well, the first part is great. I fully admit to being out of my depth when it comes to the finer points of SAS expanders, but I'm pretty sure that multipathing is why my Netapp also came with a SAS/SATA interposer card in every drive slot. I took them out because i'm only running a single path anyway and I'd rather have full access to SMART data, which SAS drives (and SAS interposers) don't give you.
|
# ? Apr 23, 2020 19:50 |
|
It's SMR time again, just some info from WD detailing the exact models, and a new post. We knew about the Reds, but even a 1TB 2.5" Black? How sad. They used to be the performance drive.. (I haven't forgotten Raptors) Western Digital posted:NOTE: All other capacity points use CMR recording technology. (Updated 4/22/2020) HalloKitty fucked around with this message at 20:30 on Apr 23, 2020 |
# ? Apr 23, 2020 20:25 |
|
HalloKitty posted:We knew about the Reds, but even a 1TB 2.5" Black? How sad. They used to be the performance drive.. (I haven't forgotten Raptors) Honestly it doesn't surprise me to see SMR pop up in 2.5" drives where space is a considerable issue, even at lower capacities. The 2 TB 3.5" Blue seems like it's a bit unnecessary, unless they're just making those as basically cut-down 6TB drives and it's therefore cheaper to make them off the same line rather than separate ones. But then it's odd that the 4TB Blue isn't listed, too, unless they just didn't get around to admitting that one yet. Still, really weird that they decided to go for SMR more deeply in the Red line than any others, despite the Red line being the most sensitive to it, rather than in the "budget" lines where you could excuse it as a simple cost-cutting trade-off.
|
# ? Apr 23, 2020 20:46 |
|
HalloKitty posted:It's SMR time again, just some info from WD detailing the exact models, and a new post. if you're buying a HD as this point why would you even believe anything they post? first they denied that they were using SMR, and then now they're releasing this info? too late. who's to say other SKUs aren't affected? what are they going to do to make their customers whole again? will they accept returns? the fact that they hid this for so long while keeping the price similar means i probably won't be buying anything from WD in the future. maybe if there was a definitive way to determine it without having to buy the drive in the first place. also the fact that their twitter account is not directly posting this in their timeline, and instead only responding when they see a tweet talking negatively about the situation goes to show how they're covering it up.
|
# ? Apr 23, 2020 20:48 |
|
Strong Sauce posted:if you're buying a HD as this point why would you even believe anything they post? first they denied that they were using SMR, and then now they're releasing this info? too late. who's to say other SKUs aren't affected? what are they going to do to make their customers whole again? will they accept returns? I hate to tell you this, but WD isn't some outlier of malfeasance in the corporate world. I would be surprised if any other manufacturer handled controversies better in any consistent way. In other words, I wouldn't write WD off just because of this because you're going to have to write off buying hard drives and basically everything made by big companies with multiple divisions that don't communicate well.
|
# ? Apr 23, 2020 21:14 |
|
Strong Sauce posted:if you're buying a HD as this point why would you even believe anything they post? first they denied that they were using SMR, and then now they're releasing this info? too late. who's to say other SKUs aren't affected? what are they going to do to make their customers whole again? will they accept returns? Hell if I know, but I'll keep an eye out for any further info from any of the manufacturers regarding this & post, since it obviously has an impact on those building NASes or any kind of storage
|
# ? Apr 23, 2020 21:25 |
|
Thermopyle posted:I hate to tell you this, but WD isn't some outlier of malfeasance in the corporate world. I would be surprised if any other manufacturer handled controversies better in any consistent way. it's a thread about storage/hard drives, not about the rest of the corporate world. i don't think they handled it very well, maybe not worse than other companies, but so what? i'm not judging them on that. i mean the reason they're doing this now is because the evidence is to the point where they can't deny it and this is to just avoid looking even worse in a lawsuit. an actual mea culpa would be to discuss how they're going to address this for the people affected now and how they'll handle this in the near future. yes, maybe in 10 years time this whole incident will have been smoothed over. but until then, not sure why i can't encourage people not to buy them or demand that they actually screw their customers? not like the SMR drives they make are cheaper. it is unfortunate that the space has been narrowed down to maybe 2-3 big players, maybe there would be more transparency in what people are buying.
|
# ? Apr 23, 2020 21:53 |
|
Strong Sauce posted:it's a thread about storage/hard drives, not about the rest of the corporate world. But, all the other hard drive companies are included in "corporate world". I'm not defending WD here, I'm saying that avoiding them in the future doesn't seem like a particularly promising way of not falling victim to the same sort of behavior. On the other hand you can think you're punishing them as an example to other companies...I just don't think it's a lesson that is going to get learnt. You should just make the best decision you can on the evidence at hand when you're ready to buy a hard drive.
|
# ? Apr 23, 2020 22:22 |
|
I don't get the angle of picking on WD because they happened to be the first ones discovered doing this, since all the other brands have now confessed they're doing it too. You're going to punish WD by... taking your business to another brand that did the exact same thing as them?
|
# ? Apr 23, 2020 22:46 |
|
|
# ? Jun 5, 2024 19:34 |
|
Paul MaudDib posted:I don't get the angle of picking on WD because they happened to be the first ones discovered doing this, since all the other brands have now confessed they're doing it too. You're going to punish WD by... taking your business to another brand that did the exact same thing as them? The difference here is WD has confirmed that they dorked up a line of drives explicitly designed and marketed for NASes with technology that almost by design doesn't play well with NASes. I think people would care a lot less if it turned out they'd slipped it into their Green line or something that is clearly marked as "mass storage the cheapest way we can make it." The other manufacturers did pretty much that--put it in low-performance consumer drives that everyone expected to be cost-cutting options in the first place, rather than in their NAS-centric offerings.
|
# ? Apr 23, 2020 23:55 |