March 28, 201412 yr Hi Guys, so yesterday after three years the thumbdrive of my unRaid Server died. The LED only glims and if i connect it to my other PC's it is recognized but with 0MB capacity and unaccessable. I just found that out because my server couldnt boot anymore. Of course i have no backup of it (how dumb). So what is the normal procedure? Of course i have to buy a new thumbdrive and then send Tom a mail so he sends me a Plus key that fits the new drive. But how it goes on from there? Is there anything special to consider when reassining the drives, apart from beeing sure which one is the parity drives? Will my user shares come back or do i have to create them all again?
March 28, 201412 yr Assuming your ONLY issue is the failed flash drive (i.e. it wasn't destroyed by a failing power supply or motherboard and the surge didn't take out any of your hard drives) ... then it's very simple. Just download the latest version of UnRAID (v5.0.5); copy all the files to your new flash drive; and run the MakeBootable script. Then boot to it; note the GUID, and send Tom a note outlining what happened. When you get a new Plus key for it, add that to the drive, and boot again. Now you simply assign all of your drives and Start the array. IMPORTANT: If you aren't CERTAIN which drive is the parity drive (and the cache drive if you had one), then only assign data drives. Then Start the array and see if any drives show as "Unformatted". Do NOT format them -- if a drive shows as unformatted, it was NOT a data drive, so start over -- do a New Config (on the Utils tab); and assign them again until you have all the data drives properly assigned. At that point, you should be able to access the array with no problem. NOW Stop the array and assign your parity drive -- then Start it again and let it do the initial parity sync. When that's done, do a parity check to ensure all went well ... and then you're done If you enable user shares, all of your shares will automatically show -- any top-level folder on a drive is treated as a share.
March 28, 201412 yr Author Thanks for the info. Thats lets me calm down a bit. The parity drive shouldnt be a problem, it is the only 3TB and the only Seagate drive in my system. So it should be easy to identify. I think it was really only the Thumbdrive which failed. Seams that they are not meant to be powered on continuously over a couple of years. What do you think, would it make sense to buy a bigger drive (eg 16GB) to reduce the wear, despite from that i dont know what really failed in my drive?
March 28, 201412 yr I don't think there's any difference in the "wear and tear" based on the size of the drives. I have 3 flash drives that have been powered on continuously since the servers were built -- 6 years in one case, 4 on another, and 2 on the last one. They're all 4GB units and are still going strong. Clearly you won't have any problem assigning the correct drive as parity, since it's the only 3TB drive
March 28, 201412 yr Author Any brand you would recommand. Althought that must not mean it lasts longer. My last drive was a 4gb Transcend.
March 28, 201412 yr The bigger the flash drive the longer it will last - as the writes to it are spread out over the entire drive. A 2G vs a 8G will last up to 15 times longer.... Myk I don't think there's any difference in the "wear and tear" based on the size of the drives. I have 3 flash drives that have been powered on continuously since the servers were built -- 6 years in one case, 4 on another, and 2 on the last one. They're all 4GB units and are still going strong. Clearly you won't have any problem assigning the correct drive as parity, since it's the only 3TB drive
March 31, 201412 yr Author Thanks for all the info. After i have received the new License Key, now everything is working like normal again.
March 31, 201412 yr The bigger the flash drive the longer it will last - as the writes to it are spread out over the entire drive. A 2G vs a 8G will last up to 15 times longer.... Myk I don't think there's any difference in the "wear and tear" based on the size of the drives. I have 3 flash drives that have been powered on continuously since the servers were built -- 6 years in one case, 4 on another, and 2 on the last one. They're all 4GB units and are still going strong. Clearly you won't have any problem assigning the correct drive as parity, since it's the only 3TB drive Do you have links to back that up? The theory is very sound but USB flash drives dont have wear leveling built in like SSD and is it really that common for people to say their flash drive is dead due to too many bad blocks... people post when it just wont work at all. Neither of these match the theory that larger USB last longer. I hope you are right but I dont think you are.
March 31, 201412 yr I'm fairly certain flash drives do not do wear-leveling like SSDs ... so a larger unit won't be any more reliable than a small unit, given the same usage pattern (i.e. an erasure/rewrite isn't going to use a different location to spread out the wear). Note that there is NO limit on reads, however, so typical UnRAID-as-a-NAS usage will likely never wear out the flash unit. If, on the other hand, you're a developer who does a lot of writes; or a user with some add-ons that write a lot of data, then you may indeed eventually have issues.
March 31, 201412 yr Some flash drives do mark bad sectors. In theory, a larger flash drive that tracks bad sectors will last longer but will get smaller over time.
March 31, 201412 yr Some flash drives do mark bad sectors. In theory, a larger flash drive that tracks bad sectors will last longer but will get smaller over time. True, if they mark bad segments, then that should allow a somewhat longer life. But I think the reliability and lifespan is already VERY good ... and it's simple enough to just contact Tom for a replacement key when one fails, so the "hassle factor" is very low. Especially if you keep a spare flash drive (a 2nd license is only $30) -- in which case it'd just be a matter of booting to the spare.
March 31, 201412 yr I suppose it matters in what quality and brand you invest in. To say that USB keys and SD cards do not do wear leveling is not true. I remember years ago reading that Sandisk and Kingston did wear leveling. Quote from an old product spec PDF I have. Wear Leveling Wear leveling is an intrinsic part of the erase pooling functionality of cards in the SanDisk SD Card Product Family using NAND memory. To say that the size of the key doesn't matter in the longevity is also not true. From personal experience, SD cards that experience allot of writes and are larger have lasted me longer. Yes, I waste a 32GB sdhc card for unRAID. That's OK. I got it on a deal and a half. <grin>
March 31, 201412 yr I'm fairly certain flash drives do not do wear-leveling like SSDs ... They support wear leveling, I do not believe they support TRIM, which is why SSD's in the right operating systems are more reliable these days.
March 31, 201412 yr I dont get it. I understand the technology behind the ideas presented here but almost every flash drive I have ever seen reported anywhere as broke has been in a failed state and not "failing". Sure some flash drives might have some wear leveling but I am prepared to bet in the grand scale of things its rare. Even if it isnt why do we rarely/ever see people saying their unRAID keys are failing as opposed to reports of failed which are common. Are we missing a trick here in early detection or at the very least boot device recommendations? Do we have anything beyond anecdotal evidence?
March 31, 201412 yr I dont get it. I understand the technology behind the ideas presented here but almost every flash drive I have ever seen reported anywhere as broke has been in a failed state and not "failing". I do not believe there is early detection to the operating systems on the USB based flash drives. The wear leveling is internal to the controller that manages the memory chip (at least from what I read). This does not get propagated to the operating system until it's too late (At least in my failures). In my case, I would just get failed writes and errors. I'm not sure that smart data can actually be accessed over USB yet. Sure some flash drives might have some wear leveling but I am prepared to bet in the grand scale of things its rare. I think it's more common in brand name sd cards then rare. It's probably not advertised as often as SSD's. Just think how many failures would be occurring if there wasn't some logic to spread around the writes. Everytime you updated a superblock or FAT you would be writing to the same spot all the time. Flash manufactures would be out of business if you could not rely on writes to either succeed or report a failure.
March 31, 201412 yr The one flash drive I had problems with did in fact exert signs of "failing" without actually being dead. It's been 3-4 years, so I don't recall the exact timings ... but the drive started taking MUCH longer to boot than it had originally. I thought perhaps it was a bad USB port, so I tried it on another computer, which had the same symptoms. It was "working" ... but an UnRAID system that had booted in perhaps 30 seconds or so was now taking 5 minutes. I tried a different flash drive with UnRAID Basic, it booted QUICK ... so I sent Tom a note and he graciously swapped my key to the new drive.
April 1, 201412 yr I find this very interesting as I might actually have a "failing" USB myself. I put it down to a USB port but maybe it isnt. Whats particularly interesting is that what we are really saying here is that we have 2 recommendations when choosing an ideal USB drive for unRAID and longevity: 1. As large as possible. 2. Preferably with wear leveling (this one is tricky as there seems to be a load of marketing FUD on this) How did we go so long without making these recommendations? Did we simply overlook it or are the improvement so intangible as to be not worthy of recommendation?
April 1, 201412 yr A failed flashdrive should not represent a major concern. I have never had one fail on its own, but have helped one along by snapping it off! Most users have a backup key to be able to recover immediately, and has been mentioned, Tom has been quick to issue replacement keys on request. Reassembling the array is easier if you have a screenshot of your current array configuration. But even if you don't, initially assigning your disks to drive slots and looking for the "unformatted" one will help you find parity. With a little help from the Trust Parity procedure, putting Humpty Dumpty back together again in pretty painless. In my test array I do some unRaid development task (myMain and preclear) and have ugdated files hundreds of times. I will not be altering my usage for fear of wearing it out and will just deal with it when it happens.
April 1, 201412 yr Agreed it is not a major cause for concern but that is no reason not to give users the right information for them to make the best flash drive purchase. I certainly would have purchased different USB drives if I had known this before and I am sure I am not alone in this.
April 1, 201412 yr The one flash drive I had problems with did in fact exert signs of "failing" without actually being dead. It's been 3-4 years, so I don't recall the exact timings ... but the drive started taking MUCH longer to boot than it had originally. I thought perhaps it was a bad USB port, so I tried it on another computer, which had the same symptoms. It was "working" ... but an UnRAID system that had booted in perhaps 30 seconds or so was now taking 5 minutes. I tried a different flash drive with UnRAID Basic, it booted QUICK ... so I sent Tom a note and he graciously swapped my key to the new drive. I bought a brand new (it was supposed to be anyway - from newegg so it probably was) SanCruzer Fit 16GB that demonstrated these symptoms. The only flash drive I ever had that was bad. It wasn't used with unRAID or as a boot device so I used it quite a while this way before I bought a new one. It was a pain in the ass having to wait several minutes before Windows would display its contents. I did have two Transcend SD readers die as well but they didn't have a unique ID so couldn't use them with unRAID anyway so was another minor loss.
April 2, 201412 yr The irony is not lost on me that as we speak I have been hit by the USB key "failing" gods. For reference i added these words to the hardware compatibility section of the wiki. Feel free to enhance: There is some evidence that choosing a USB flash drive much larger than you actually require will increase its life span as wear will be spread out over the larger device. It is also believed that branded USB flash drive will generally turn out to be more reliable than a generic unbranded ebay purchases. If possible choose a USB flash drive that features wear leveling as it may under certain circumstances increase device longevity (unconfirmed).
October 3, 201411 yr The one flash drive I had problems with did in fact exert signs of "failing" without actually being dead. It's been 3-4 years, so I don't recall the exact timings ... but the drive started taking MUCH longer to boot than it had originally. I thought perhaps it was a bad USB port, so I tried it on another computer, which had the same symptoms. It was "working" ... but an UnRAID system that had booted in perhaps 30 seconds or so was now taking 5 minutes. I tried a different flash drive with UnRAID Basic, it booted QUICK ... so I sent Tom a note and he graciously swapped my key to the new drive. 100% exact same situation happened on a build I did for a friend. (brand name stick but I think a plug-in back in the day was making a lot of writes to it) I very much like the idea of a card reader over a USB stick for a couple reasons. First no fear if the card dies just plug in another. Secondly, it allows you to have a quick change configuration backup incase something goes wrong. Now that has become less of a concern lately with better handling of additional programs. And yes finding a card reader with an unique guid isn't easy and yes the reader could break but I think the pros out weigh the cons. Sent from my SM-N9005 using Tapatalk
Archived
This topic is now archived and is closed to further replies.