|
UndyingShadow posted:Running 4 VMs I thought it would be nice to give them 2 (or even 3) cores each? If they don't utilize those cores they are causing more harm due to cpu scheduling (if you over allocate). What workloads do you actually plan on running?
|
![]() |
|
DNova posted:Or you could buy a good processor. Seriously. Current AMD core does not equal current (or even one or two generations old...maybe even three?) Intel core.
|
![]() |
|
UndyingShadow posted:That would be fine, except I plan to shove 8 drives into my whitebox, and that IBM only holds 4 Then check out the TS440. Quoting myself from several pages ago: SamDabbers posted:For anyone looking to build a new NAS with similar requirements, take a look at the Lenovo TS440. It comes with four 3.5" hot swap bays, and can be upgraded to eight. You'll also have to buy drive sleds, because the ones included are just dummy placeholders.
|
![]() |
|
Also, note that a hardware thread from hyperthreading does not really perform the same as an actual core even for virtual machines and can cause some strange performance characteristics here and there (scheduling 2 vCPUs to run on two separate hardware threads across two cores when they could have been on the same core at least, which gets worse in SMP situations). However, this does not mean that 8 AMD cores > 4-core hyperthreaded Intel cores when it comes to these really easy virtualization cases. Unless you're running something like a goddamn Hadoop cluster on your server, it's unlikely for you to be actually affected by any of this besides the downsides of the AMD processor. I ran Hadoop on my E3-1230 for a while with 2 slaves and a single master because 32GB of RAM was the bigger problem for my workloads, not my CPU by a long shot (and the whole point of trying to distribute jobs is to try to better distribute the CPU load around, so not much point using stuff like Hadoop or Storm in my case).
|
![]() |
|
I have a great little Freenas 9.3 setup with 7 1Tb disks in a RaidZ3 array, just used as dumb storage that I access every once in a while for media. The box is pretty quiet, but every once in a while it sounds like the disks are being accessed. Not a constant accessing type of drive grinding, but in short spurts, and I don't know why. The only services I run are CIFS and SMART. While the disks are churning I'm not accessing them, nor are any other computers in my apartment, nor is there a scrub running or any process that I am running over the disks, nor do I let anyone outside my network in to access anything such as this machine. What could this be? Does this happen to anyone else? Is this something that freenas does? All parts are brand new, and there are no errors on any of the disks, and the array scrubs clean. Anyone have ideas on what could be accessing the disks periodically?
|
![]() |
|
Moey posted:If they don't utilize those cores they are causing more harm due to cpu scheduling (if you over allocate). IOwnCalculus posted:Seriously. Current AMD core does not equal current (or even one or two generations old...maybe even three?) Intel core. Okay. AMD is out. necrobobsledder posted:Also, note that a hardware thread from hyperthreading does not really perform the same as an actual core even for virtual machines and can cause some strange performance characteristics here and there (scheduling 2 vCPUs to run on two separate hardware threads across two cores when they could have been on the same core at least, which gets worse in SMP situations). However, this does not mean that 8 AMD cores > 4-core hyperthreaded Intel cores when it comes to these really easy virtualization cases. Unless you're running something like a goddamn Hadoop cluster on your server, it's unlikely for you to be actually affected by any of this besides the downsides of the AMD processor. I ran Hadoop on my E3-1230 for a while with 2 slaves and a single master because 32GB of RAM was the bigger problem for my workloads, not my CPU by a long shot (and the whole point of trying to distribute jobs is to try to better distribute the CPU load around, so not much point using stuff like Hadoop or Storm in my case). The only thing that will be constant is a FreeNAS VM, a Pfsense VM for a 300mbps cable connection, and a windows server 2012 R2 for playing with. There will be others that come on and offline, for fun and testing. I was looking at the hyperthreading Xeons, because in my mind, I'm trying to build the best, most powerful system I can. I'm considering the SUPERMICRO MBD-X10SLM+-F-O, which has 2 nics, and an IPKVM nic (which now that I know about it, seems like a really great thing to have, as I hate lugging things around to hook up consoles.) I'm also gonna look at the TS440. When you say you have to buy drive sleds, do you have to buy all 8 drive sleds, or just the 4 for the add-on cage?
|
![]() |
|
I have TS440, it is a pretty awesome machine. Comes with an SAS controller in IT mode ( mine did anyway, i have the 1245V3 version ) and has enough space for drives. You will need caddies for all your drives, it does not come with any. They cost quite a lot in Europe, so I would buy them on Ebay or something. They can be had for like 20 bucks or something. The case is sturdy and it feels like a quality machine. I like it very much. Oh and it is fast.........
|
![]() |
|
necrobobsledder posted:However, this does not mean that 8 AMD cores > 4-core hyperthreaded Intel cores when it comes to these really easy virtualization cases.
|
![]() |
|
Combat Pretzel posted:AMDs 8-core CPUs are some mixed clusterfuck. I think each two cores are a module that share a single x86 decoder, FPU and some other things, while having their own integer ALUs. Or some shit like that. If I were to look it up, I wouldn't be surprised to find that a hyperthreaded Intel CPU would probably mop the floor with these AMD CPUs. I didn't know this. Ugh. I haven't bought an AMD cpu since the core duo days, and it looks like there's absolutely no reason to, ever.
|
![]() |
|
UndyingShadow posted:I didn't know this. Ugh. I haven't bought an AMD cpu since the core duo days, and it looks like there's absolutely no reason to, ever. Pretty much, yeah. I have a couple of their netbook-level chips (one in an actual netbook that is my garage laptop, one as my pfSense box) and otherwise there's nothing compelling about anything AMD at all. They do a better job of not restricting features based on market segmentation (i.e. ECC / VT-d support) but thanks to how common it is, you can still get your hands on a used Nehalem (or newer) Xeon with a real server board and ECC RAM for the same price as a new shitty AMD CPU and the few motherboards that officially support ECC / IOMMU.
|
![]() |
|
UndyingShadow posted:The only thing that will be constant is a FreeNAS VM, a Pfsense VM for a 300mbps cable connection, and a windows server 2012 R2 for playing with. There will be others that come on and offline, for fun and testing. Meanwhile if you got 1Gbps fiber, you can pretty much get fucked like I had to(*), if you're thinking of running pfSense without an ASIC or FGPA of some description (of which very few are actually network processors which can handle routing, NAT, ipv4+6 and ACLs that are available, at least in Denmark along with being supported in FreeBSD). *: I ended up buying an EdgeRouter Lite 3 and 3x UniFi APs with zero-handoff which have served me much better, so it worked out fine in the end.
|
![]() |
|
UndyingShadow posted:I didn't know this. Ugh. I haven't bought an AMD cpu since the core duo days, and it looks like there's absolutely no reason to, ever. --edit: Wait, I've got my shit wrong. "Steamroller" --edit2: Apparently that shit is out already. Either way, get an Intel. Combat Pretzel fucked around with this message at 18:33 on Jan 23, 2015 |
![]() |
|
Megaman posted:I have a great little Freenas 9.3 setup with 7 1Tb disks in a RaidZ3 array, just used as dumb storage that I access every once in a while for media. The box is pretty quiet, but every once in a while it sounds like the disks are being accessed. Not a constant accessing type of drive grinding, but in short spurts, and I don't know why. The only services I run are CIFS and SMART. While the disks are churning I'm not accessing them, nor are any other computers in my apartment, nor is there a scrub running or any process that I am running over the disks, nor do I let anyone outside my network in to access anything such as this machine. What could this be? Does this happen to anyone else? Is this something that freenas does? All parts are brand new, and there are no errors on any of the disks, and the array scrubs clean. Anyone have ideas on what could be accessing the disks periodically? Bump, anyone?
|
![]() |
|
Megaman posted:Bump, anyone? Pull the network plug? See if it still occurs? If it does, it is something in the OS itself (cron?). If it stops it might be a client indexing or something. I don't know, might help you pinpoint the cause.
|
![]() |
|
Mr Shiny Pants posted:Pull the network plug? See if it still occurs? If it does, it is something in the OS itself (cron?). Good point, I'll pull the plug, but let's say I do and it continues (which I guarantee it will), when you say cron you mean one that I put on? I put no crons on, this is a fresh install. Does Freenas come with crons I don't know about that should be running besides scrubs that I've scheduled? Also, when you indexing, what do you mean? Does raid/freenas require some sort of indexing? I've never heard of that before.
|
![]() |
|
Megaman posted:Good point, I'll pull the plug, but let's say I do and it continues (which I guarantee it will), when you say cron you mean one that I put on? I put no crons on, this is a fresh install. Does Freenas come with crons I don't know about that should be running besides scrubs that I've scheduled? I am not familiar with Freenas so I am not sure if it creates certain cron jobs or other scheduled tasks. I can imagine that it does for reporting data usage or some other data aggregation. Seems like it can: http://doc.freenas.org/9.3/freenas_tasks.html Maybe it has some system defined tasks that get kicked off at regular intervals and those tasks access your disks? As for indexing, I meant a Windows system that indexes your files for searching or spotlight in OSX. As I said before, I am not familiar with Freenas, but I can imagine that these settings show up as the behavior you describe.
|
![]() |
|
D. Ebdrup posted:Do note that running between 100 and 500Mbps through pfSense requires at least 2GHz CPU on a modern Intel CPU (since AMD has been ruled out previously) according to the hardware requirements. I run pfsense on a celeron g1620 and have 1Gb fiber. I can get it to max out my bandwidth but I don't use it to vpn.
|
![]() |
|
Megaman posted:I have a great little Freenas 9.3 setup with 7 1Tb disks in a RaidZ3 array, just used as dumb storage that I access every once in a while for media. The box is pretty quiet, but every once in a while it sounds like the disks are being accessed. Not a constant accessing type of drive grinding, but in short spurts, and I don't know why. The only services I run are CIFS and SMART. While the disks are churning I'm not accessing them, nor are any other computers in my apartment, nor is there a scrub running or any process that I am running over the disks, nor do I let anyone outside my network in to access anything such as this machine. What could this be? Does this happen to anyone else? Is this something that freenas does? All parts are brand new, and there are no errors on any of the disks, and the array scrubs clean. Anyone have ideas on what could be accessing the disks periodically? Mine does this occasionally, even when it's not being touched. I think it's just part of FreeNAS. Disk IO is gonna happen. Just enjoy the soothing sounds of your hard drives and why worry?
|
![]() |
|
I'm looking to replace my current homebrew NAS case with something a bit better because I'm sick of it's low build quality. What's out there with 24 drive bay (or more) capability? I had a look at Backblaze storage pods but not a fan of not having drive bays to be honest.
|
![]() |
|
UndyingShadow posted:Mine does this occasionally, even when it's not being touched. I think it's just part of FreeNAS. Disk IO is gonna happen. Just enjoy the soothing sounds of your hard drives and why worry? Here's the system activity on my SSD: ![]()
|
![]() |
|
One more question. I've just ordered a whole new set of 7 4TB disks to replace my existing 7 1TB disks. I read the documentation on how to replace disks, and I feel like the FreeNAS documentation although extensive, kinda sucks sometimes, it's a little vague, or maybe I'm just stupid. How do you replace a disk in FreeNAS 9.3? Do you just pull a disk to replace, insert the new disk, and wait until FreeNAS automatically resilvers it, then move onto the next disk? Or do I actually have to tell the disk I'm replacing to deactivate before removing it, then tell it to use the new disk when putting it in? How does replacing a disk work exactly? FreeNAS documentation appears to imply that it's all automatic, and all I need to do is just pull then insert. Ideas?
|
![]() |
|
mayodreams posted:I run an almost identical setup to what you are looking to build. If you would like to do esxi, I would strongly recommend you get a board with a supported chipset (opteron/xeon) from VMware's HCL. You're running a virtualized router? Have you had any issues with that setup? I've been thinking about switching ISPs and I'd go from 30/3 to 100/5. I'd think this would keep up possibly. Do you have a recommended setup/parts list?
|
![]() |
|
RE: replacing disks in RAIDZ vdevs under FreeNAS You go to the ZFS zpool under Storage, go to View Status, and from there you Offline a disk to mark the disk ID as not participating and to free up a device, replace the disk physically, then you go back to the UI and hit Replace. This is how you'd do it under ZFS via command line although once before years ago on OpenSolaris I'm not sure what happened and ZFS automatically offlined a disk for me and after I replaced it ZFS recognized the new disk and the vdev started to get resilvered. Maybe I'm completely crazy though because I likely did it at 3 am and everything blurred together.
|
![]() |
|
I have a problem where I'm trying to put some bigger drives into my linux server, but the system isn't seeing them. I had a 750gb drive and it was visible, and then I replaced it with a WD Red 3TB and the system doesn't see it. I've already got a bunch of them in the system, some on the onboard controller, some on my M1015. lshw shows the missing drive is hooked up to the M1015 but I'm not sure what else to look at. I've got the drives in a 5-in-3 enclosure, could it be a power issue? Anything else I should try and look at?
|
![]() |
|
What's the preferred cloud backup service around these parts? Crashplan? Backblaze? I have about 2-3TB of data I would like not to disappear in the unfortunate event of a house fire or flood.
|
![]() |
|
Crashplan. Lisa needs backups.
|
![]() |
|
Crashplan, yet again.FISHMANPET posted:I have a problem where I'm trying to put some bigger drives into my linux server, but the system isn't seeing them. What do you mean it can't be seen? Does it not have a /dev/sdx assigned to it as a logical name when it shows up in lshw?
|
![]() |
|
Nope, there's no indication that the drive is physically plugged in. lshw shows 7 drives connected to my M1015, when I in fact 8 drives. parted only shows 12 physical drives when I have 13. There's no /dev/sdx entry. Just, nothing. I'm not sure if there's any other ways for me to try and probe the controllers. I guess I should also look at the POST messages to see if it shows up there.
|
![]() |
|
Oh, I read your first post as if it was showing up in lshw. Assuming your M1015 is in IT mode, there's nothing else to do beyond just plugging a drive in. I'd plug it into another box just to rule as much as possible out, but DOA is always a possibility.
|
![]() |
|
This happened to me before, when I tried to get my two raid sets to each live nicely in their own enclosure. (I had 5x3TB and 5x1.5TB plus 2x250gb and an empty slot then) I had to mix them up to get them to work. So it's not a dead drive issue, could there be some kind of power issue? I don't know how much more power a 3TB drive takes over some other older drive.
|
![]() |
|
Hate to beat a topic to death, but why does FreeNAS "require 8GB" of RAM and Nas4Free suggest an "minimum of 1GB with ZFS"? I'm thinking of buying a N54L which comes with 4GB RAM... replacing it with 8GB is +~$90 on a $180 machine, which is a harder sell. Going to be using this a light-duty office storage server, primarily as a backup target and some other light interoffice file sharing stuff for 4 users. Somewhere between 2 x 2TB and 3 x 3TB of storage. Will 4GB be a performance killer for a relatively light usage NAS? My old synology at home doesn't seem to have problems pushing nearly gigabit speeds with 256MB RAM. Is ZFS that RAM hungry?
|
![]() |
|
drk posted:Hate to beat a topic to death, but why does FreeNAS "require 8GB" of RAM and Nas4Free suggest an "minimum of 1GB with ZFS"? I'm thinking of buying a N54L which comes with 4GB RAM... replacing it with 8GB is +~$90 on a $180 machine, which is a harder sell. It's a best-practices kind of thing. Without compression and deduplication, 4GB of RAM will probably be completely fine. edit: I don't like the idea of 2x2TB. Get at least 3x1TB in RAIDZ-1 (total usable space 2TB). I'd recommend a 4-drive RAIDZ-2. edit: Ok 4-drive RAIDZ-2 might be overkill for you, but get at least 3 drives in a RAIDZ-1. sleepy gary fucked around with this message at 22:44 on Jan 26, 2015 |
![]() |
|
4GB of RAM will be fine for 4-6 TB of usable space in a office use setting.
|
![]() |
|
DNova posted:It's a best-practices kind of thing. Without compression and deduplication, 4GB of RAM will probably be completely fine. Thanks for the thoughts. Any reason a 3-drive RAID Z1 is preferable to a simple 2 drive mirrored setup? If price was no object, it would obviously be a more efficient use of space. I'm trying to pitch the idea to the new job I am working at, where their idea of backing up is: 1) nothing or 2) dragging around a shitty external USB drive when they remember to do it. Of course, the external option is probably going to die before the computers it is backing up, since those drives are almost universally terrible. Since the N54L has 4 drive bays, I am also assuming it is easier to expand storage needs in the future by adding 2 drives to 2, versus adding 1 drive to 3, though I haven't worked out exactly how I'd go about that.
|
![]() |
|
drk posted:Thanks for the thoughts. Any reason a 3-drive RAID Z1 is preferable to a simple 2 drive mirrored setup? If price was no object, it would obviously be a more efficient use of space. I'm trying to pitch the idea to the new job I am working at, where their idea of backing up is: 1) nothing or 2) dragging around a shitty external USB drive when they remember to do it. Of course, the external option is probably going to die before the computers it is backing up, since those drives are almost universally terrible. If you need to spend as little as possible now and have an eye towards expansion in the near future, then a 2-disk mirror is maybe the way to go. Later, you can add another mirror of 2 disks to the pool. It's not a great solution but it's better than a shitty external drive that is going to get dropped at some point.
|
![]() |
|
was looking at NAS options over the past few days; currently I'm using my main PC to serve files which is... less than optimal especially since all my media, photos etc is on two 5 year old shitty seagates in RAID-0. originally was going to build some ridiculous thing using a DS380, xeon etc but I think I'll start small since my use case for something like that is... shaky at best. heard good things about the hp microservers - the newest model (gen8) seems to have an upgradable cpu and looks a little nicer, is there anything majorly wrong with it that I should know before I pull the trigger? would it also be possible to put a low end i3/xeon in it sometime in the future in case i wanted to do a plex transcode or two? without causing a house fire
|
![]() |
|
Generic Monk posted:all my media, photos etc is on two 5 year old shitty seagates in RAID-0. Why would you do this? Why? Why? quote:heard good things about the hp microservers - the newest model (gen8) seems to have an upgradable cpu and looks a little nicer, is there anything majorly wrong with it that I should know before I pull the trigger? would it also be possible to put a low end i3/xeon in it sometime in the future in case i wanted to do a plex transcode or two? without causing a house fire Yes lots of people have put Xeons in their Gen8 Microservers. Here is one example: http://homeservershow.com/hp-prolia...pu-upgrade.html
|
![]() |
|
DNova posted:Why would you do this? Why? Why? Also note that these are the Barracuda 7200.11 drives that shipped with a firmware defect that caused them to randomly drop out of arrays. They did this about 3 times before I just used a win8 storage space (sigh) to patch together something that seemed to work. At this point I'm just not fucking with it until I have a more permanent solution, everything irreplaceable is in a dropbox or dropbox-like anyway. Also that's great; nice to know I have my options open - as open as the ebay market for discontinued CPUs is anyway. Pulled the trigger, though I see myself spending twice the cost of the bare server on drives and RAM anyway ![]()
|
![]() |
|
Generic Monk posted:heard good things about the hp microservers - the newest model (gen8) seems to have an upgradable cpu and looks a little nicer, is there anything majorly wrong with it that I should know before I pull the trigger? would it also be possible to put a low end i3/xeon in it sometime in the future in case i wanted to do a plex transcode or two? without causing a house fire Damn, that's a nice deal on that box.
|
![]() |
|
I just decided to upgrade my ancient Dlink DNS323 after nearly 8 years of solid, but very slow use. Synology 215j should arrive tomorrow. Does anyone know if I have to reformat the drives I'm using when I put them in the Synology?
|
![]() |