|
frogbs posted:So I just bought a synology DS209, and am having a few issues with it. I'm accessing it from OSX using AFP. Most of the time its fine, transfers are fast and it seems to be funcitoning perfectly. Randomly, the volume will just crash and will have to rebuilt (im running a raid 1 on two seagate 1tb drives). It seems to do this completely randomly, although I have noticed that it does it more often when I am copying from a SMB share to it, although I doubt thats related. Is my configuration to blame, or is it more likely that there is a problem with one of the drives or the synology itself? I don't have that model but have had no issues with AFP/SMB/iSCSI and Synology. What version of DSM are you using? RAID1 should be reliable...
|
#
?
Dec 13, 2010 20:27
|
|
|
|
| # ? Apr 26, 2024 05:38 |
|
|
kri kri posted:Is anyone using Greyhole? If so what are your impressions? Looking to move my 20 or so disk fileserver over to greyhole in the very near future. Been testing it out in a virtual setup and it seems to work just fine. It doesn't really do anything weird, so if bad comes to worse, you can easily pull your files from the pool-disks.
|
|
#
?
Dec 13, 2010 21:16
|
|
|
Scuttle_SE posted:so if bad comes to worse, you can easily pull your files from the pool-disks. That seems like one of the great features of greyhole. You're not dependent upon it at all when it comes to data recovery.
|
|
#
?
Dec 13, 2010 22:15
|
|
|
what is this posted:I don't have that model but have had no issues with AFP/SMB/iSCSI and Synology. What version of DSM are you using? RAID1 should be reliable... I'm on the latest version, DSM 3.0-1354.
|
|
#
?
Dec 13, 2010 23:44
|
|
|
I'm using DSM3 as well. I'd have to vote for drive problems, particularly since you're rebuilding the array and it's dead-simple RAID1. Is it the same drive that goes bad every time?
|
|
#
?
Dec 14, 2010 00:27
|
|
|
I'd appreciate it of someone could give some input to a setup I've been pondering... on a purely performance note, would I gain anything by moving my webserver+minecraft server data (which is a collection of chunks, 10's of thousands of files, lots of i/o...) from my 6x1tb (7200.12) raidz1 to a raid0 pair of PATA 300s? Just how stupid does that sound with pata...? Would also be nice to not bother the media pool when the minecraft server is running, so room-mate can pull/push to it without fear of compromising server performance. Currently on a dual-core openindiana box, possibly quad core come christmas.
|
|
#
?
Dec 14, 2010 01:53
|
|
|
Wanderer89 posted:I'd appreciate it of someone could give some input to a setup I've been pondering... on a purely performance note, would I gain anything by moving my webserver+minecraft server data (which is a collection of chunks, 10's of thousands of files, lots of i/o...) from my 6x1tb (7200.12) raidz1 to a raid0 pair of PATA 300s? Just how stupid does that sound with pata...? Would also be nice to not bother the media pool when the minecraft server is running, so room-mate can pull/push to it without fear of compromising server performance.
|
|
#
?
Dec 14, 2010 01:59
|
|
|
adorai posted:I think you would be better off with zil/arc. 2 pata 300's will have pretty lovely performance compared to even your existing raidz1. That's what I figured, sata dual-platter 1tb's had to be faster than some aging 300 ide's... Thanks for pointing me at zil/arc, I'll look into it.
|
|
#
?
Dec 14, 2010 02:07
|
|
|
paradigmm posted:Same deal here. Volume kept crashing. I had enough with it and got this: Jeez anyone know how to get the US sale price in Canada US Price http://www.newegg.com/Product/Product.aspx?Item=N82E16859321014&Tpk=acer%20aspire%20easystore http://www.newegg.ca/Product/Product.aspx?Item=N82E16859321014CVF&cm_re=acer_Aspire_Easystore-_-59-321-014CVF-_-Product I'd buy immediately at the US Price of 350, but the CDN price of $500 (keep in mind that the exchange rate has been hovering around 96-99 cents for the past few months. It's not quite the dual GB thing I needed the synology for in the office but it's exactly what I need for home.
|
|
#
?
Dec 14, 2010 03:08
|
|
|
what is this posted:I'm using DSM3 as well. It never tells me that a drive is bad, it just tells me that the volume has crashed and then it re-verifies the array.
|
|
#
?
Dec 14, 2010 03:52
|
|
|
Likely a drive is finding a bad sector and going into heroic recovery mode, which makes it stop responding to commands for about two minutes while it tries to recover the data in that sector. Drives designed for RAID won't do this; if they find a bad sector and it takes more than a few seconds to recover, they just drop the sector and report they're bad drives. Taking so long means they will report they are just fine, but the RAID controller or RAID logic will drop the drive for being unresponsive for so long. Check the SMART bad sector counts to verify. If you've got sector reallocations, you have a drive that is likely to die and is literally on its way out.
|
|
#
?
Dec 14, 2010 04:33
|
|
|
Factory Factory posted:Check the SMART bad sector counts to verify. If you've got sector reallocations, you have a drive that is likely to die and is literally on its way out. Ignorant person here. I thought it was to be expected that some drives would have bad sectors and that it was only a problem if you started seeing a lot (hundreds? thousands? bajillions?) of them.
|
|
#
?
Dec 14, 2010 04:36
|
|
|
frogbs posted:It never tells me that a drive is bad, it just tells me that the volume has crashed and then it re-verifies the array. In Linux the RAID device breaks the volume whenever an IO operation fails. This would suggest you should see a lot of relocated sectors in the SMART report if it is a disk issue, otherwise something fruity with the controller. Sometimes this is a hardware failure, sometimes I guess it could be a drive incompatibility with the NAS controller.
|
|
#
?
Dec 14, 2010 04:53
|
|
|
Thermopyle posted:Ignorant person here. I thought it was to be expected that some drives would have bad sectors and that it was only a problem if you started seeing a lot (hundreds? thousands? bajillions?) of them. Google put out a whitepaper on consumer drive failure (since that's what they use in their server farms). Basically, 2+ reallocated sectors drops a drive's 1-year survival from ~97% to ~60%, with more sectors meaning a worse prognosis.
|
|
#
?
Dec 14, 2010 05:00
|
|
|
Scuttle_SE posted:Looking to move my 20 or so disk fileserver over to greyhole in the very near future. Been testing it out in a virtual setup and it seems to work just fine. It doesn't really do anything weird, so if bad comes to worse, you can easily pull your files from the pool-disks. Which OS will you be using? I am considering a move from whs v1 I would be curious as to how your migration goes.
|
|
#
?
Dec 14, 2010 06:57
|
|
|
kri kri posted:Which OS will you be using? I am considering a move from whs v1 I would be curious as to how your migration goes. In this case, I'm going to go with either Debian or Ubuntu...leaning towards Ubuntu, it's more updated
|
|
#
?
Dec 14, 2010 11:21
|
|
|
I have a question that is best suited to the expertise in this thread! I hope you can give me some advice! I currently have a mdadm Raid 5 setup with 5x500GB discs running on Ubuntu I have two spare SATA connections on a controller card and I am wondering what my best move is to essentially upgrade my Raid. I am looking to purchase some 2TB drives in place of the 500's. What are my options here? Ideally I want to keep Raid 5, is there any way for me to swap the drives out for larger ones and keep the raid running? Sorry I will admit I have a limited knowledge of the mechanics of what I can and can't do with software Raids Thanks, Sorry if my question is unclear or has been worded oddly!
|
|
#
?
Dec 16, 2010 16:22
|
|
|
GhostSeven posted:I have a question that is best suited to the expertise in this thread! I hope you can give me some advice! You can create a "degraded" RAID5 array. This means that you could put two 2TB drives on the two ports you have available and then copy data off of one of your current drives onto that array, remove that 500GB drive, put in another 2TB drive, and add it to the degraded RAID5 array. Something like this: "mdadm --create --level=5 --raid-devices=3 --force /dev/md0 /dev/sda /dev/sdb /dev/sdc missing" That creates the array in the same state as it would be if /dev/sdc had failed. This means the array is offering you no protection until you add the third 2TB drive. Disclaimer: I haven't done this, do some more research before you try it. I do think it would work, though. (Or you could spend 30 bucks on a cheap PCI-SATA adapter and add more ports)
|
|
#
?
Dec 16, 2010 18:03
|
|
|
Thermopyle posted:You can create a "degraded" RAID5 array. This means that you could put two 2TB drives on the two ports you have available and then copy data off of one of your current drives onto that array, remove that 500GB drive, put in another 2TB drive, and add it to the degraded RAID5 array. Oh that is quite a fantastic idea, I am assuming mdadm would not mind if after the copy I removed the 500GB drives and reshuffled the 2TB ones and then added a drive, mdadm generally being quite smart and sorting that stuff out. I will do some more research on this but thanks! Edit -- Had a new thought. If I popped out my boot drive thus giving me three total free SATA ports and then booted from a live cd / usb stick could I build the new Raid, load up the old Raid, copy the data across then pop the drives out shuffle around and boot back up? Or is ubuntu likely to have a fit? - To clarify the boot drive is just a bog standard 80gb SATA drive. GhostSeven fucked around with this message at 18:19 on Dec 16, 2010 |
|
#
?
Dec 16, 2010 18:12
|
|
|
Is your end goal a 5x2TB RAID, or a 7x2TB RAID?
|
|
#
?
Dec 16, 2010 18:22
|
|
|
Hi guys, I currently have a hacked-up WDTV Live + 1TB USB HDD acting as a NAS to my home theater setup. Performance sucks (5MB/s reads over the wire) and I'm almost out of space. I've been thinking of getting a larger/most robust setup and it seems like my best bet would either be an Atom WHS server (Acer Easystore D510 4TB for ~$540 incl. an additional 2TB drive) or a Synology DS211 (~$500 incl. 2 separate 2TB drives). I intend to have it serve content to my Boxee Box as well as my PC. It has to be able to handle torrents and nzb downloads while doing the prior. I'm pretty OS agnostic but am inclined to going with Synology right now considering that the Acer will be obsolete once Vail comes out. Power consumption is pretty important as I've calculated that a 100W difference will save me $200 a year. What do you guys think? shrike82 fucked around with this message at 23:15 on Dec 16, 2010 |
|
#
?
Dec 16, 2010 23:12
|
|
|
FISHMANPET posted:Is your end goal a 5x2TB RAID, or a 7x2TB RAID? I think my goal initially would be 3x2TB with room for expansion which was how my 500GB one started had 3x500 and then moved up to 5x500. Should I be thinking about maybe ZFS though I am not sure how well that runs on Ubuntu. I would move to a different OS but this is a jack of all trades box. It runs my internet and XBMC out to my TV
|
|
#
?
Dec 17, 2010 09:15
|
|
|
ZFS runs like a dog, even more so in Ubuntu through ZFS-Fuse, to be honest. But that said, "like a dog" for me means 80 MB/s read/write in sustained transfers while giving me snapshots and double redundancy on four 2TB drives. You mostly pick ZFS for the features, not the raw performance. And if you don't need the features, stick with what works. Random and small-file I/O performance will probably be the worst, but that may be acceptable for a set-top box.
|
|
#
?
Dec 17, 2010 10:54
|
|
|
Okay, I think I have decided on things. This is going to be my first NAS setup, it's main purpose will be streaming 1080p media to my Boxee Box and occasionally my desktop/netbook. I'm looking at getting the Synology DS211j and placing 2 WD 1.5TB Green Drives in it. Any advice?
|
|
#
?
Dec 18, 2010 00:58
|
|
|
Has anyone pulled the trigger and tried out Solaris Express 11? I've got a new fileserver box sitting next to me and I'm hemming and hawing over OpenIndiana and full-on Solaris. It seems it's still too new for any reasonable comparison articles to have been written. Edit: Ha! Nevermind... apparently Oracle doesn't give access to patches/updates unless you're paying. That makes it an easier choice, I guess! devilmouse fucked around with this message at 01:47 on Dec 18, 2010 |
|
#
?
Dec 18, 2010 01:43
|
|
|
devilmouse posted:Has anyone pulled the trigger and tried out Solaris Express 11? I've got a new fileserver box sitting next to me and I'm hemming and hawing over OpenIndiana and full-on Solaris. Mine is running Express 11. The thing with no patches though, is not as big a deal as you think. This is a really old, really stable code base. Most bugs are very very unique corner cases that affect very specific setups.
|
|
#
?
Dec 18, 2010 04:40
|
|
|
Wasnt there a post from the ZFS guys where they didn't guarantee rebuilds on vdevs larger than 8-9 drives? A 22 disk vdev is just retarded. You need more vdevs to get the iops performance.
|
|
#
?
Dec 18, 2010 16:08
|
|
|
Anybody had problems with consumer NAS products that boast Time Machine capability/compatibility? I just bought a LaCie 5big Network 2 NAS, and I'm having no luck at all with its (supposed) Time Machine feature. I enabled the NAS's global Time Machine functionality, created a new share for TM, and enabled the specific TM protocol for that share. When I hit "Select Backup Disk..." in TM preferences on the Mac, my TM share is nowhere to be found. It actually worked, once, but then after that it never worked again so I started over from scratch and since then I haven't been able to get the share to even show up in the list of candidate backup disks. Also, the LaCie Network Assistant application doesn't seem to work on the Mac, either. Windows machines on my network can run it and see the NAS no problem, but the Mac version always just says "LaCie networks disks cannot be found." I can connect to the share from the Mac via "Connect to Server..." from the finder, and I can connect to the NAS's web-based configuration dashboard from the Mac's web browser (just browsing to the local IP of the NAS), but the stupid LaCie application that is supposed to make dealing with the NAS easier refuses to work at all. Anybody own this particular NAS, or had similar issues with another?
|
|
#
?
Dec 19, 2010 00:31
|
|
|
conntrack posted:Wasnt there a post from the ZFS guys where they didn't guarantee rebuilds on vdevs larger than 8-9 drives? A 22 disk vdev is just retarded. The iops of a vdev is equal to the iops of the slowest individual drive in the vdev. With a vdev of that size you're sacrificing general speed, but you're also rolling the dice on your mean time to data loss. If you have a disk go in a 22 disk raidz, it's basically got to read the data off all 22 disks to reconstruct the missing disk.
|
|
#
?
Dec 19, 2010 08:55
|
|
|
A little more experience from the zfs-fuse under Ubuntu front. I benchmarked 80 MB/s reads and writes using dd, but in real-world performance, the disks would behave oddly, spending half their time writing and half their time waiting. I started doing real-world-like tests with fio, and it was really clear that I was topping out at 47 MB/s reads and averaging around 37 MB/s. After some Googling and another thread, I tried disabling prefetching, which was as simple as entering the following command to restart the daemon: code:It's instantly repeatable, since the filesystem is in userspace. Just re-enter zfs-fuse with no arguments to re-enable prefetching, and it slows to a crap again. Now, there was a trade-off. With prefetching, random read tests ran at a ludicrous 200 MB/s. Now it's an extremely crappy average of 783 KB/s. But I can live with that for a single user fileserver pool. To make the prefetch disable permanent, I just edited /etc/zfs/zfsrc to uncomment a single line. Simple. For reference, the pool exists on four Samsung Spinpoint F4 2 TB disks, partitions aligned but I don't know as ZFS clusters are. A RAID10 set in the first 60 GB of each drive benches at 250 MB/s max 210 MB/s average sequential read.
|
|
#
?
Dec 19, 2010 13:18
|
|
|
Will the Synology DiskStation DS411J take harddrives of different sizes at the same time? I can only confirm that it maxes out at 4 x 2TB.
|
|
#
?
Dec 19, 2010 19:10
|
|
|
zfs-fuse is not ready for prime time. Starting the daemon with the computer is buggy and sometimes fails, requiring a reinstall of the entire package. Not worth the hassle, here comes Solaris Express 11.
|
|
#
?
Dec 20, 2010 04:26
|
|
|
Lilleput posted:Will the Synology DiskStation DS411J take harddrives of different sizes at the same time? I can only confirm that it maxes out at 4 x 2TB. The rules that it has are the same that Linux mdraid has. So, no, it's not Drobo style. Most people prefer it that way. If you need to hang a little extra storage off it you can always just JBOD them up inside, or plug it in via USB until it's time to reload with fresh drives.
|
|
#
?
Dec 20, 2010 20:29
|
|
|
Jonny 290 posted:The rules that it has are the same that Linux mdraid has. So, no, it's not Drobo style. Most people prefer it that way. Actually, the current firmware offers drobo style mixed hard drive sizes quote:Synology Hybrid RAID (SHR) http://www.synology.com/enu/products/features/Volume_Management.php That said I don't recommend using a Drobo or Drobo-style "hybrid RAID" or "unRAID" or whatever you want to call it. I simply don't trust any of these systems and there's always a performance hit. Slower, and less reliable? Where do I sign up?
|
|
#
?
Dec 20, 2010 21:25
|
|
|
Learn somethin new erry day! Neat. Still wouldn't use it, though, like you said.
|
|
#
?
Dec 20, 2010 21:59
|
|
|
what is this posted:Slower, and less reliable? Where do I sign up? Do you have any evidence that it is less reliable other than the obvious "less time in production than regular raid"?
|
|
#
?
Dec 20, 2010 23:58
|
|
|
Not synology's, no. It's too new for me to know about the reliability (hint: that means don't trust it). Their other stuff has been very solid in my experience, and they have some of the better software for consumer NAS decives. There are countless drobo horror stories out there however. And I do mean countless. I've experienced drobo disasters in person.
|
|
#
?
Dec 21, 2010 00:04
|
|
|
what is this posted:Not synology's, no. It's too new for me to know about the reliability (hint: that means don't trust it). Their other stuff has been very solid in my experience, and they have some of the better software for consumer NAS decives. So...the answer is "no, unless you're talking about Drobo".
|
|
#
?
Dec 21, 2010 03:04
|
|
|
Drobo's been the main company selling devices that do this. Unless you count Windows Home Server, which had a bunch of issues early on with data corruption, and now dropped the feature from the upcoming release because of issues with data corruption and horrible slowdowns in heavy/enterprise usage (admittedly it was fine in small consumer setups). Hard drives are extremely cheap and in two years you can buy a new rack of drives. You may even want a new NAS. You can expand a RAID set with existing drive sizes without issue and without using the unevenly sized drives faux-raid feature. The only reason to want different sized drives is because you have a bunch of junky old hard drives lying around, maybe a 250GB drive here, a 500GB drive there, a 1.5TB drive there, and hey just throw out the old small drives and buy a few 1TB, 1.5TB, or 2TB drives and be done with it. Hard drives are really, really cheap. Giving up speed and reliability just because you have a four year old 250GB drive sitting in a USB enclosure that you think you can save some money on to store your precious animes is a hilarious joke. Buy a drive 10 time the size for $100.
|
|
#
?
Dec 21, 2010 03:29
|
|
|
|
| # ? Apr 26, 2024 05:38 |
|
|
So uh, what would I use and how could I use it to map physical drives to a VM?
|
|
#
?
Dec 21, 2010 03:46
|
|
