|
Flipperwaldt posted:I would think so yeah. You can't put extra ram in either, so 2gb is a way better starting point. There will be a point when one can get the newest software and the other won't, but currently it's only the 2012 and earlier models that can't update to dsm 7, so that's still a good while away for the 218 (longer than the lifespan of the disks you'd put in now). I can't really think of a positive reason to go for the 220j if it isn't maybe the price being half that of the other. If the 220+ were in your price range, that would be a worthwhile upgrade cpu wise, but I'm assuming you've looked at that. That’s what I thought, I was just surprised that they hadn’t really updated the hardware in a couple years. I have a DS218+ myself, this one will be for my parents who do decidedly less nerdy things with it so I can’t justify the extra cost. I was looking for a DS220j for them and stumbled across a DS218 for only $20 more, and I’d imagine the DS218 would be supported at least as long as the DS220j just based on specs alone.
|
#
?
Oct 16, 2022 13:42
|
|
|
|
| # ? May 15, 2024 11:51 |
|
|
Welp when I set up a truenas scale install I set up redundancy on everything but the boot SSD and guess which drive is the one that decided to fail. But I'm posting this to say it actually worked very well. I got a notification that boot-pool was degraded due to too many checksum errors so I exported the config, ordered a new drive (I still need a spare but I want to see if I can warranty the original first), rebooted it once to verify it wasn't a weird one-off thing and sure enough got an I/O error during boot, shut the server down, unplugged the old drive, put the new one in, reinstalled truenas on it, booted that, imported the old config, and everything's back to the way it was. Whole thing took less than hour once I had the new drive in hand and the system ran completely fine with the failed boot drive until then.
|
|
#
?
Oct 16, 2022 14:40
|
|
|
power crystals posted:Welp when I set up a truenas scale install I set up redundancy on everything but the boot SSD and guess which drive is the one that decided to fail. But I'm posting this to say it actually worked very well. I got a notification that boot-pool was degraded due to too many checksum errors so I exported the config, ordered a new drive (I still need a spare but I want to see if I can warranty the original first), rebooted it once to verify it wasn't a weird one-off thing and sure enough got an I/O error during boot, shut the server down, unplugged the old drive, put the new one in, reinstalled truenas on it, booted that, imported the old config, and everything's back to the way it was. Whole thing took less than hour once I had the new drive in hand and the system ran completely fine with the failed boot drive until then. Dunno if truenas allows you to automate the config export but you'll probably want to setup a job to do that once a day/week and include it in your offsite backup plan.
|
|
#
?
Oct 16, 2022 14:43
|
|
|
Rexxed posted:The upside of those kind of disks is that they're small, but the downside is that internally it's a thick laptop style 2.5" drive. They make them like that because they can be entirely powered by the one plug instead of an external power brick like a 3.5" disk and they're fairly cheap. They're also kind of slow just as drives. Thank you! Yeah, I have an off-site automated backup as well, this is just extra protection incase that fails. I guess i'll go for it, it seems as good an option as any. One weird thing, the 5tb drive is only $5 more than the 4tb. Should I stay away from 5tb drives for some reason?
|
|
#
?
Oct 16, 2022 17:15
|
|
|
frogbs posted:One weird thing, the 5tb drive is only $5 more than the 4tb. Should I stay away from 5tb drives for some reason? Nah, just pricing variance. External drive sales are like that. Both the 4tb and 5tb drives are using SMR tech. With some types of backup methods SMR can be very slow. Any time that you're overwriting chunks of old data rather than writing new data SMR is not great. But I don't know how TimeMachine organizes data, so I'm not sure whether that will be an issue or how bad it might be. You might ask in the Mac thread about this, someone there might know. (Though if the bulk of your backed-up data is relatively static & isn't frequently changing, ex a big collection of music or photos, it probably doesn't matter either way. Your backups only get updated when the files change.)
|
|
#
?
Oct 16, 2022 17:46
|
|
|
Is there any/some benefit in writing syslog in Truenas SCALE to the zfs data pool instead of OS disk? Is there any/some benefit in having system dataset (samba permissions and encryption keys etc) in zfs data pool instead of OS disk? It's just that writes related to that stuff are so small they accumulate, and zfs writes the accumulated data to disk every 5s. And that makes some extra repeating noise 12 times a minute. A SSD would be better place for that. I bought these HDD's really mainly for very large backups. I assume that even if the Truenas scale OS installation gets vaporized I can just add the HDD's to a new truenas install, and it figures based on the data on disk that they are an existing pool and recovers it without deleting any data on the disk. (Or by restoring a backed up truenas config file.) Ihmemies fucked around with this message at 20:33 on Oct 16, 2022 |
|
#
?
Oct 16, 2022 17:50
|
|
|
I'm putting logs and poo poo on the boot-pool. Dedicating a 1TB SSD to torrents, apps/VMs and other poo poo that slams the drives with random read/write. Don't need a bunch of random io hitting spinning rust. Finished transferring 44TB last night, that only took most of the weekend  Apparently rsync is CPU limited and topped out at 160-180MB/s per thread on my temporary setup using an i5-2500K. Managed to run 2-3 threads for most of it, but half of the dataset was one folder, so that took the longest. One of these days I'll learn how to use rclone.
|
|
#
?
Oct 17, 2022 05:57
|
|
|
frogbs posted:Thank you! Yeah, I have an off-site automated backup as well, this is just extra protection incase that fails. I guess i'll go for it, it seems as good an option as any. I don't know of any reason why it would be better or worse. These things usually have a certain amount of GB per platter so it might be 4 vs. 5 1TB platters spinning or something. Maybe they've got the 2.5" ones with density between 1 and 1.5 TB or something, but I'm not a drive manufacturing expert, I mostly take them apart when they die. Either way, they should be similar internally, just don't move it while it's spinning and try not to drop it.
|
|
#
?
Oct 17, 2022 08:42
|
|
|
My 18TB 3,5" Toshibas have 9 platters. So that's 2TB/platter, 1TB/side. Are those 2,5" disks SMR or how do they fit all that inside such a small space?
|
|
#
?
Oct 17, 2022 14:16
|
|
|
Ihmemies posted:My 18TB 3,5" Toshibas have 9 platters. So that's 2TB/platter, 1TB/side. Are those 2,5" disks SMR or how do they fit all that inside such a small space? Very much SMR. A while ago the 1 & 2tb 2.5" drives were CMR, back when WD published their whole list of drives after every one got mad. But today I dunno.
|
|
#
?
Oct 17, 2022 15:55
|
|
|
My new NAS idles at 134W  That's with a Xeon E5-2670 v3, 128GB RAM, Quadro P400, intel 10GbE card (SFP+), LSI 9211 8i, 8x14TB (shucked WD Elements, so 7200RPM) in raidz2 and a couple of SSD's. I'm very happy with that.
|
|
#
?
Oct 17, 2022 19:44
|
|
|
I'm surprised you can get by, pulling only 134W on that hardware. 
|
|
#
?
Oct 17, 2022 20:21
|
|
|
Wibla posted:My new NAS idles at 134W 134, Really? I Don't want to piss in your cornflakes, but that seems too low
|
|
#
?
Oct 17, 2022 21:57
|
|
|
Wibla posted:My new NAS idles at 134W Hell yeah. I wish the IPMI on my Supermicro board would spit out power consumption, cause I don't want to have to shut my whole rack down to put a kill a watt on it. Your build is pretty similar to mine (2620v3, 70GB RAM, P2000, 9211, 10GbE Mellanox, 8x 8TB drives in 2 vdevs and some cache SSDs) so this is a good datapoint to have.
|
|
#
?
Oct 17, 2022 22:01
|
|
|
Adolf Glitter posted:134, Really? I Don't want to piss in your cornflakes, but that seems too low Could be with the data drives spun down, maybe?
|
|
#
?
Oct 17, 2022 22:05
|
|
|
Crunchy Black posted:Hell yeah. I wish the IPMI on my Supermicro board would spit out power consumption, cause I don't want to have to shut my whole rack down to put a kill a watt on it. A good UPS should offer at least a total wattage figure for whatever you've got on it, obviously not granular to the machine unless you have a REALLY nice UPS but it's probably a safe bet what the majority of the consumption is from. My UPS shows 138 watts right now for a load mostly made up of my server (Ryzen 1600, 32GB RAM, Geforce 970, Mellanox 40G, LSI 8 port, 8x12TB WD shucked) and a 24 port UniFi non-PoE switch which seems about right in line.
|
|
#
?
Oct 17, 2022 22:08
|
|
|
wolrah posted:A good UPS should offer at least a total wattage figure for whatever you've got on it, obviously not granular to the machine unless you have a REALLY nice UPS but it's probably a safe bet what the majority of the consumption is from. Good idea!  There's a cloud key, USG, Switch 24, Switch 8 150 and a H3C 10GbE PoE switch along with 3 RPis so that is probably reasonably accurate.
|
|
#
?
Oct 17, 2022 22:22
|
|
|
Adolf Glitter posted:134, Really? I Don't want to piss in your cornflakes, but that seems too low That's with the drives spinning, measured at the wall.
|
|
#
?
Oct 17, 2022 22:26
|
|
|
Crunchy Black posted:Good idea! That's not even close to accurate according to what I'm actually measuring and what that comes up as. If you switch to the status tab it will give you how many amps it supplying. V * I = P 123.1volts * ?amps = ?watts In my case the way you did it comes up as 645 when using the measured amperage (which is in agreement with the PDU) I get 484.
|
|
#
?
Oct 17, 2022 22:36
|
|
|
My setup wouldn't be great for a NAS, but I could really use a DAS, especially since I'm already a backblaze user. What should I be looking at? I wouldn't mind hw raid 5 combined with offsite cloud mirror..all drobo stuff seems like it's totally sold out, qnap looks like it has a decent 4 bay option..just wondering what the best AIO DAS is
|
|
#
?
Oct 17, 2022 23:29
|
|
|
zer0spunk posted:My setup wouldn't be great for a NAS, but I could really use a DAS, especially since I'm already a backblaze user. Are you sure you want a DAS? Those don't do any form of RAID, they're typically fairly dumb boxes that just let you connect more drives to a controller. That controller can then do hardware RAID, but that's sort of separate. Assuming you accidentally swapped NAS and DAS, the thread seems to mostly use Synology for AIO boxes and TrueNAS on whatever hardware if you need something more custom - but others know more about those than me.
|
|
#
?
Oct 18, 2022 00:19
|
|
|
Computer viking posted:Are you sure you want a DAS? Those don't do any form of RAID, they're typically fairly dumb boxes that just let you connect more drives to a controller. That controller can then do hardware RAID, but that's sort of separate. My network bandwidth isn't great, and I'm not planning on serving any other devices. Having a poolable attached storage device means it's all consolidated properly, can be on or off independent of the pc, and can still be backed up to my cloud choice which allows attached storage but not a NAS obviously I just don't have the use case for a NAS with my setup, a DAS looks like the move
|
|
#
?
Oct 18, 2022 00:30
|
|
|
Edit: beaten on the NAS/das thing I’m guessing they want direct attached given everything else they’ve said, especially the Backblaze thing. Promise, OWC, and Sandisk makes some I’ve seen people use; OWC uses some kind of softraid I know, not sure about the others. I have no personal experience other than Drobo though.
|
|
#
?
Oct 18, 2022 00:31
|
|
|
Oh yeah, I've actually used an Icybox USB thing that took four disks and had a RAID mode that presumably would present them as a single large drive. I only used it with the untouched drives to expand a ZFS pool, but I should have remembered that this is a category that absolutely exists. Apologies.
|
|
#
?
Oct 18, 2022 00:34
|
|
|
I thought about maybe just doing individual drives pooled together with drivepool instead of raid 5's 3 drives + parity and then using snapraid for said parity instead..but the problem there is that the spreading of files would make the offsite backblaze backup a real pain in the rear end if I had to rely on that. Having to figure out what files were actually on the bad drive to restore doesn't seem worth the hassle. I could rely on just snapraid, but then it's down to raid is not a backup again. If I go hardware controller raid 5 + backblaze cloud of that array I'd feel much better. The real irreplaceable stuff like important document scans would all go on 250 gig sata ssd, encrypted, which would get backed up on the raid, in the cloud and also placed offsite at a friend's place.
|
|
#
?
Oct 18, 2022 01:35
|
|
|
The real difference between a DAS and a NAS is whether the computer the disks are attached to has a network file sharing deamon running. 
|
|
#
?
Oct 18, 2022 10:28
|
|
|
BlankSystemDaemon posted:The real difference between a DAS and a NAS is whether the computer the disks are attached to has a network file sharing deamon running. This popped up on HN and I was curious about what you'd say. https://taras.glek.net/post/curious-case-of-maintaining-sufficient-free-space-with-zfs/ Obviously the general wisdom is "don't try to run at the absolute limit of your pool" and I assume the same applies to dataset quotas. Does a scrub trigger release of any pending blocks? It could probably refcount at that point right? Or is this more analogous to dedup where just data integrity is checked and it's not doing a refcount? (why can't dedup do a refcount during a scrub and free unused dedup blocks anyway? does it not walk the actual snapshot tree, just the data?)
|
|
#
?
Oct 23, 2022 23:08
|
|
|
Paul MaudDib posted:This popped up on HN and I was curious about what you'd say. That being said, you need to be careful about space considerations on every filesystem. ZFS and UFS solved this by having reservations - for ZFS it's a per-dataset property that can be set anywhere, and for UFS it defaults to 8% of the filesystem at creation, that can only be written to by root once all other space has been used. Obviously this doesn't work for ZFS, because 8% of hundreds of TB is quite a lot of diskspace to reserve. Also, in particular, point 3 that the blogger makes is easily solved nowadays by having a script that automatically ages snapshots out by converting them to bookmarks. That's because zfs-bookmark(8) is a way of keeping a txg for a snapshot without the prior records being referenced - meaning you can still do incremental replication pulled to an offsite that has all the snapshots (because you can't create them on offsite, as they'll get overwritten by subsequent zfs send|receive). A scrub reads every instance of every record (ie. in case of ditto blocks, mirroring or striping with distributed parity) and compares it with its checksum, and that's all it does if memory serves. The delete operation in ZFS is a background task that's asynchronous as has been observed, and if memory serves there's some reasons for this buried in the code comments surrounding its implementation, but the gist of it is that it's the only way to do things properly. It's up to the operator to ensure sufficient space - by implementing proper (non-email based) notifications through monitoring. BlankSystemDaemon fucked around with this message at 07:49 on Oct 24, 2022 |
|
#
?
Oct 24, 2022 07:40
|
|
|
Adolf Glitter posted:134, Really? I Don't want to piss in your cornflakes, but that seems too low I just checked my dual Xeon E5-2690 v3 + 8 HDD system and it was idling at ~200 watts. Oof. Think I'm gonna have to let it hibernate for the winter, possibly even replace the two Xeons with a single low power chip.
|
|
#
?
Oct 24, 2022 15:06
|
|
|
This isn't NAS, but it is pretty storage-y. I'll be building a new pc in a couple days, and I was considering rolling some windows Storage Spaces / Pools on it, to use some older harddrives without as much worry that they'll suddenly kick the bucket. (The 2 drives I'll be reusing from the old pc are around 6 or so years old. It seems more practical to take the hit to capacity and keep using them, than retiring them early just in case.) Is it really as easy as just firing up powershell to dedicate some SSD or NVME as cache, pairing it up with some HDDs in the pool, and off you go? And are there any huge gaping issues I should be mindful of? Up until now I've been using Primocache to do some SSD caching of drives, and I've been pretty happy with the result, and I was hoping spaces/pools could be a decent way of replicating the performance boosts, while also having some simple redundancy on these drives. (Especially since I have a mix of older, smaller ssds where I could just dedicate a whole one to caching and capacity for a pool.) Setup would probably be something like Performance pool: 6+6+8 TB, where the 2 6s are old drives, with 2-way mirror plus an old, small nvme drive as cache. Mostly project files and etc, stuff I have backed up, but primarily large things I don't want clogging up my SSDs. Bulk pool: A new 12TB drive paired with a different old ssd as cache, just to make it a bit more snappy. I have backups and etc of anything important, so it's just redundancy in the sense that it won't be too annoying if one of the old drives kick the bucket in the near future... just as long as both don't do it at the same time. Unrelatedly: I don't suppose there's a resource anywhere that keeps track of what drives have obnoxiously loud PWL?
|
|
#
?
Oct 24, 2022 15:12
|
|
|
My storage server thing (based on 2012 essentials) has been running DrivePool for its whole existence with no issues. It was incredibly easy to setup and I've never had to touch it since. I just used it for replication, not performance, since it's effectively my NAS (or was until I recently decided to replace it).
|
|
#
?
Oct 24, 2022 17:12
|
|
|
Smashing Link posted:I just checked my dual Xeon E5-2690 v3 + 8 HDD system and it was idling at ~200 watts. Oof. Think I'm gonna have to let it hibernate for the winter, possibly even replace the two Xeons with a single low power chip. It looks like the E5-2650L V3 is a 65 Watt TDP part that would be a lower clock speed but same core count Xeon for that socket. The 2690 v3 is 135 Watt. There's some on ebay starting around $40. When I deal with older Xeons I like to refer to this list on wikipedia: https://en.wikipedia.org/wiki/List_of_Intel_Xeon_processors_(Haswell-based)#Xeon_E5-2690_v3
|
|
#
?
Oct 24, 2022 21:49
|
|
|
Rexxed posted:It looks like the E5-2650L V3 is a 65 Watt TDP part that would be a lower clock speed but same core count Xeon for that socket. The 2690 v3 is 135 Watt. There's some on ebay starting around $40. That's a great resource, thanks for sharing. Looks like the energy savings would pay for the chip in about 4 months. Would be sad to put the old boys out to pasture though. 
|
|
#
?
Oct 25, 2022 15:46
|
|
|
I ended up going with a Synology DS220+ to replace my QNAP 2bay (5 years old). Its night and day better. The only downside is its more annoying to install some apps, you need to ssh in and run them via docker and give them permissions to folders. It's a one time setup cost so its fine. It was even easy to mount in the folders from my old NAS to copy everything over. The box uses very little CPU/RAM and has been very snappy with the same processes running as my QNAP. I saw you can upgrade to 16GB of ram but also that not all RAM is compatible, does anyone have a link to one that will work?
|
|
#
?
Oct 25, 2022 16:26
|
|
|
The soldered on ram has D4NESO printed on it and as a consequence that code will show up in many listings for compatible sticks. I added 4GB myself and shopping for it was confusing, with reviews for the stick both claiming it would and wouldn't work. And then it didn't work with the nas fresh out of the box (wouldn't boot). And then it did work adding it after I had the whole thing set up and I'm obviously not sure at all why. Shop somewhere with a good return policy. Look for specific mentions of the ds220+, not just any Synology device. If a specific sku for a ram stick is mentioned, you want that one and not anything equivalent. I can look up what the 4GB one was I used if you want. 16GB, no clue. This reddit thread might be your best bet for candidates.
|
|
#
?
Oct 25, 2022 17:50
|
|
|
I set out to make my setup energy efficient, my entire network, servers, switches, APs, cameras, everything, idles at 220w.
|
|
#
?
Oct 25, 2022 17:53
|
|
|
Smashing Link posted:That's a great resource, thanks for sharing. Looks like the energy savings would pay for the chip in about 4 months. Would be sad to put the old boys out to pasture though. E: 220 watts idle here for network stuff and my NAS box with dual 2696v4s and 2080S.
|
|
#
?
Oct 26, 2022 09:26
|
|
|
Kivi posted:Look for any low core count v4 instead. Should be drop in placement with higher efficiency, due to smaller manufacturing process. Looks like I can get a 2560L v4 for $50...may just give it a try. I'm also going to go from 8 HDDs to probably 3 so that may help a bit.
|
|
#
?
Oct 26, 2022 18:34
|
|
|
Holy poo poo people, are you all running home labs or something? I average that for total power consumption over a month, including an oven, and fish tank among other things.
|
|
#
?
Oct 26, 2022 18:40
|
|
|
|
| # ? May 15, 2024 11:51 |
|
|
mobby_6kl posted:Holy poo poo people, are you all running home labs or something? I average that for total power consumption over a month, including an oven, and fish tank among other things. uhh, how? most people have been talking about 200-300w system draw, the average electric oven is about 3000w. Hell, even heaters for fish tanks go up to 1000w (I've got too many fish tanks, one with a 1000w heater and several in the 200-300w range)
|
|
#
?
Oct 26, 2022 18:54
|
|
