«608 »
  • Post
  • Reply
D. Ebdrup
Mar 13, 2009



H2SO4 posted:

I just bought another seagate backup plus 8TB from Amazon to fully populate my DS1019+. I've ran FreeNAS forever but i have been impressed by the synology - it's like a grown-up Drobo.
It's a grown-up Drobo in that it'll lose your data and tell you it's its fault, rather than blaming you?

BeastOfExmoor
Aug 19, 2003

I will be gone, but not forever.


Nam Taf posted:

Grabbed 6 of the 10TB drives last night when they were still on sale. They state shipping on 3/6. Pray for me that it doesn't get cancelled.

I'm hoping you're in a country where 3/6 means June 3rd and not March 6th?

Red_Fred
Oct 21, 2010



Fallen Rib

Red_Fred posted:

Is anyone using Synology Surveillance and DS Cam? I got it all setup but I’m not convinced that DS Cam and my phone are working as a geofence to turn home mode on and off.

I can’t properly connect to the camera when I’m not on my wifi which is a good sign but I’m also not getting any notifications so say it’s turning on and off.

Ok I got it sorted, you need to use your QuickConnect name not the IP. However the geofence still seems to be problematic.

Pardot
Jul 25, 2001


I remember reading some article several years ago that looked at the error rate of drives and it argued that the larger the drives you use the more likely you'll get fucked on a restore. Has anything changed there? It'd be nice to grab a lot of 10TB drives to replace my current 4TB, but it seems risky.

Axe-man
Apr 16, 2005

The product of hundreds of hours of scientific investigation and research.

The perfect meatball.


Clapping Larry

As a general rule of thumb most RAID failures happen during a restore or parity recovery. So that might be what it is refering too.

gary oldmans diary
Sep 26, 2005



Hell Gem

Can I power a several external hard drives with this or would that be a regrettable mistake?
https://www.amazon.com/dp/B076HKCRNG

Rexxed
May 1, 2010

Dis is amazing!
I gotta try dis!



gary oldmans diary posted:

Can I power a several external hard drives with this or would that be a regrettable mistake?
https://www.amazon.com/dp/B076HKCRNG

You'll have to look at the power output of the transformers they use now. If each drive uses about 1A @ 12V then that could theoretically handle five of them. I wouldn't trust the current rating on a generic power supply like that and would assume it's going to handle less than it says it will. I wouldn't trust my data to that rating.

Chumbawumba4ever97
Dec 31, 2000



gary oldmans diary posted:

Can I power a several external hard drives with this or would that be a regrettable mistake?
https://www.amazon.com/dp/B076HKCRNG

I personally wouldn't use something like that. Use the original power supplies and this: https://www.amazon.com/dp/B004LZ5XM...i_1I84CbPV3VMSB

It's not as "neat" but it's way more safe for your drives.

Paul MaudDib
May 3, 2006

"Tell me of your home world, Usul"


Pardot posted:

I remember reading some article several years ago that looked at the error rate of drives and it argued that the larger the drives you use the more likely you'll get fucked on a restore. Has anything changed there? It'd be nice to grab a lot of 10TB drives to replace my current 4TB, but it seems risky.

Basically, every drive has an "Unrecoverable Bit Error" rate at which they are expected to maintain data integrity. Consumer drives are rated at one bit error per 10^12 reads, while enterprise drives are 10^14 or something like that. Those numbers haven't improved as much over time and so by the letter of the math, there is a pretty high probability of at least one error cropping up during a rebuild of a >10 TB array.

In practice though, drive reliability tends to vastly exceed the official rating. If that rating were real then you would hit errors all the time during a ZFS scrub, and I've never seen even one over ~100 scrubs on my dozens-of-tb pools. In practice, a scrub is a "dress rehearsal" for a resilver operation since all data needs to be read out of the pool to validate the checksums anyway.

Also, a sane soft-raid should just lose/corrupt that file instead of failing the whole rebuild over a single bit-error.

Pardot
Jul 25, 2001


Paul MaudDib posted:

Also, a sane soft-raid should just lose/corrupt that file instead of failing the whole rebuild over a single bit-error.

Thanks, this makes sense. I looked a bit, but cant find for sure if the synology hybrid raid does this. Does anyone know?

necrobobsledder
Mar 21, 2005
Lay down your soul to the gods rock 'n roll

Nap Ghost

Errors in batches of drives have fairly high cohort based correlation. What that means is that drives from a similar batch of drives from the same place and time tend to fail together. This also leads to the likelihood of failures during a rebuild higher because the drives probably came together as a group when some idiot was drunk and dropped the whole pallet of them, some wind blew onto the line in Thailand, etc.

Atomizer
Jun 24, 2007

Bote McBoteface. so what


gary oldmans diary posted:

Can I power a several external hard drives with this or would that be a regrettable mistake?
https://www.amazon.com/dp/B076HKCRNG

I think the last time I had an external HDD plugged into my KillAWatt it measured an average draw of ~10 W, from a PSU rated for 20 W (likely to handle the surge upon spin-up.) Since P=IV (or W = A * V) that thing you linked is rated for 60 W. It might be able to support a few drives in active operation, but wouldn't be able to handle the surge of all 8 drives spinning up at once. I wouldn't try it.

H110Hawk
Dec 28, 2006


gary oldmans diary posted:

Can I power a several external hard drives with this or would that be a regrettable mistake?
https://www.amazon.com/dp/B076HKCRNG

I would get one rated higher, but unlike the nay-sayers here I don't see much of a problem with it. Inrush current on small 5400rpm motors isn't that much, and it should be leveled off by some capacitors on the boards themselves. I would still make sure everything is correctly backed up. 120->12v transformers have been a thing for a long time. Try not to buy the cheapest one.

MagusDraco
Nov 11, 2011

even speedwagon was trolled


necrobobsledder posted:

Errors in batches of drives have fairly high cohort based correlation. What that means is that drives from a similar batch of drives from the same place and time tend to fail together. This also leads to the likelihood of failures during a rebuild higher because the drives probably came together as a group when some idiot was drunk and dropped the whole pallet of them, some wind blew onto the line in Thailand, etc.

I can say I got bit by this in two different cases (kinda). Ordered 4 drives at the same time from the same place and they all had similar production dates (8TB WD Reds) half of them were DOA the other two died 1-2 years later. This could have been someone drop kicking newegg's box though since one drive was "spindle was broken and making grinding noises, drive won't even show up in the bios dead"

earlier this year I ordered a single drive (8TB Seagate Ironwolf) and noted the production date. It failed within amazon's return period so I got a replacement. The replacement's production date was within a couple weeks of the first failed drive and it ALSO failed in the same way (reallocated sector count going up). The replacement for that replacement's production date was several months later and it's been fine so far. Hope it lasts at least a few years.

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!



BeastOfExmoor posted:

I'm hoping you're in a country where 3/6 means June 3rd and not March 6th?
Yes I live in what’s commonly referred to as “literally anywhere but the US”.

In any case, the orders were cancelled. I got 2x $30 vouchers out of it, but burned my prime trial memberships so that sucks. It looks like they all got cancelled, judging by others’ replies.

Heners_UK
Jun 1, 2002


Nam Taf posted:

In any case, the orders were cancelled. I got 2x $30 vouchers out of it, but burned my prime trial memberships so that sucks. It looks like they all got cancelled, judging by others’ replies.

This calm and pensive response has no place in a tech-deals discussion! Unless the hard drives are delivered to you pre-shucked on a gold platter, you and all others should fly off the deep end and say that Amazon built it's entire business over many years just to screw you and only you at this and only this moment.

CommieGIR
Aug 22, 2006

If Godzilla can do it, you know I can deliver!

Pillbug

I was having issues with FreeNAS throughput and it wasn't liking my disk controller on my SuperMicro board, so I got bored and did something.

I virtualized FreeNAS on my Xen cluster. Installed MDAM on the Xenserver, RAID-5'ed the 8 1TB drives together, then broke those into 8 500GB clusters in Xen to the the FreeNAS VM, and it actually is getting BETTER throughput than it did on metal.

Only major downside is my external USB 3.0 backup drive can only passthrough as a USB 2.0, so rsync is slow between the cluster and the USB, but overall I'm really happy with how it came out, and Xen gives me better use of the hardware, and management over the box resources. Plus I can use some of the RAID-5 cluster for handling snapshots and backups of my other VMs

I went overboard I know, and I was skeptical about the performance of the Array being RAID'ed then broken back up into a ZFS pool, but I can't find anything to complain about.

CommieGIR fucked around with this message at 22:47 on May 22, 2019

gary oldmans diary
Sep 26, 2005



Hell Gem

Nam Taf posted:

Yes I live in what’s commonly referred to as “literally anywhere but the US”.
You completely forgot all the places Y/M/D is used.

Schadenboner
Aug 15, 2011

I MEAN, TURN OFF YOURE MONITOR, MIGTH EXPLAIN YOUR BAD POSTS, HOPE THIS HELPS?!

Anyone who uses anything other than YYYY-MM-DD has opinions which are literally poop from a butt (the butt is they mouf).

D. Ebdrup
Mar 13, 2009



CommieGIR posted:

I was having issues with FreeNAS throughput and it wasn't liking my disk controller on my SuperMicro board, so I got bored and did something.

I virtualized FreeNAS on my Xen cluster. Installed MDAM on the Xenserver, RAID-5'ed the 8 1TB drives together, then broke those into 8 500GB clusters in Xen to the the FreeNAS VM, and it actually is getting BETTER throughput than it did on metal.

Only major downside is my external USB 3.0 backup drive can only passthrough as a USB 2.0, so rsync is slow between the cluster and the USB, but overall I'm really happy with how it came out, and Xen gives me better use of the hardware, and management over the box resources. Plus I can use some of the RAID-5 cluster for handling snapshots and backups of my other VMs

I went overboard I know, and I was skeptical about the performance of the Array being RAID'ed then broken back up into a ZFS pool, but I can't find anything to complain about.
I can't see how it would be faster unless there's some caching going on, but unless you're running the Xen and do a lot of tracing, I'm not sure how you can find out.

edit: Actually, since you mention Xen and I assume you don't run Xen on FreeBSD, I think it may be because Linux might still treat all disks as block devices instead of character devices, and thus do some form of caching.
FreeBSD got rid of this back in 1999 for what seems like excellent reasons.

Schadenboner posted:

Anyone who uses anything other than YYYY-MM-DD has opinions which are literally poop from a butt (the butt is they mouf).

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!



gary oldmans diary posted:

You completely forgot all the places Y/M/D is used.

Which still interpret a two-item date as dd/mm in my experience. I’ve only ever had it be ambiguous to Americans.

gary oldmans diary
Sep 26, 2005



Hell Gem

Nam Taf posted:

Which still interpret a two-item date as dd/mm in my experience. I’ve only ever had it be ambiguous to Americans.
If other countries are going out of spec on their own standards () for the sake of putting least significant figures first ( again) then it sounds like America finally has a system of units worth emulating. No point in switching to a system for consistency when that system itself is internally inconsistent.

H2SO4
Sep 11, 2001

put your money in a log cabin




Buglord

CommieGIR posted:

I virtualized FreeNAS on my Xen cluster. Installed MDAM on the Xenserver, RAID-5'ed the 8 1TB drives together, then broke those into 8 500GB clusters in Xen to the the FreeNAS VM, and it actually is getting BETTER throughput than it did on metal.

What kind of demon are you trying to summon? The first commandment of ZFS is "present your disks directly to ZFS" and you're sticking like three layers of abstraction in the middle.

D. Ebdrup
Mar 13, 2009



That's always been the mystery of ZFSonLinux to me: How does it cope with the kernels device caching?

CommieGIR
Aug 22, 2006

If Godzilla can do it, you know I can deliver!

Pillbug

H2SO4 posted:

What kind of demon are you trying to summon? The first commandment of ZFS is "present your disks directly to ZFS" and you're sticking like three layers of abstraction in the middle.

It was a test to see how it'd work. Frankly, so far so good. The unknown is why I've got good full backups.

I just got sick of fighting FreeNAS over controllers. Like I said in the first post, it was a stupid, crazy idea that I wanted to try. So far so good.

CommieGIR fucked around with this message at 21:54 on May 23, 2019

Alzabo
Oct 23, 2002

You watched it, you can't unwatch it.


I have a spare Intel x79 system laying around, converting it into a FreeNAS box doesn't seem like a bad idea?

Can the thread recommend a HBA that's gonna support 8 direct attach WDred or shucked WD drives, and where I can buy it?

Crunchy Black
Oct 24, 2017

CASTOR: Uh, it was all fine and you don't remember?
VINDMAN: No, it was bad and I do remember.




Alzabo posted:

I have a spare Intel x79 system laying around, converting it into a FreeNAS box doesn't seem like a bad idea?

Can the thread recommend a HBA that's gonna support 8 direct attach WDred or shucked WD drives, and where I can buy it?

Basically any LSI/Avago/Broadcom that is in "IT mode"

e: if you held a gun to my head, here you go: https://www.ebay.com/itm/New-LSI-Me...XAAAOSwdGFYwCX-

H2SO4
Sep 11, 2001

put your money in a log cabin




Buglord

That looks like the next generation of the 2008 based model I've been running in a few boxes for years. Not sure if 12Gb bandwidth would actually buy you anything with consumer drives though. I run the LSI 9211-8i which you can find for like $75 on amazon.

Paul MaudDib
May 3, 2006

"Tell me of your home world, Usul"


Alzabo posted:

I have a spare Intel x79 system laying around, converting it into a FreeNAS box doesn't seem like a bad idea?

X79 is perfectly fine as long as you will tolerate the power. It's sub-Ryzen performance but HEDT power consumption. Being able to run ECC RDIMMs/cheap 128GB+ capacity is cool, and that's something that's unique to X79 (unless you go with a 2011-3 server board) but personally I would go with something else these days. The merits of ECC RDIMMs have faded as the memory crisis has failed.

  • If you want to go low-power with a validated ECC platform, go with a 7100 and a C236 board. Or wait for 9300 to really hit the market and go with a C246 board. SuperMicro says 9-series BIOS support will be coming on C246 "later this year" (although that's vague)
  • If you want maximum storage or networking, in a small chassis, SuperMicro has boards with up to 8x SATA3 + 8x SAS3 in a mATX format. Or they make variants with 10GbE onboard, I think maybe even a variant with both?
  • Or wait for Asrack to get their X470D4U board validated on Zen2 (the Intel boards have better SATA/expansion/GPU transcoding support though)
  • If you want maximum expansion, go with an E5-1660v3 and a X99 WS-E or a X99 WS/IPMI. So far it's still the best option for 7 full-size PCIe slots or 5+IPMI with 40 lanes to back it up, unless you jump to dual-socket.
  • If you want maximum processing power, go with a Threadripper board or wait for Zen2.

Alzabo posted:

Can the thread recommend a HBA that's gonna support 8 direct attach WDred or shucked WD drives, and where I can buy it?

IBM M1015 flashed into IT mode, it's got an LSI controller underneath and it's widely used in the white-box NAS community. Or if you want to have 16 ports per card, LSI makes a card for that too (16i/16e suffix is 16 ports internal or external, I want to say 9300-16i?).

And yeah LSI is your best bet as far as compatibility. You do need to watch driver support on FreeBSD, and FreeNAS is not always up to date with the underlying FreeBSD either.

Paul MaudDib fucked around with this message at 19:01 on May 24, 2019

bobfather
Sep 20, 2001

I will analyze your nervous system for beer money

If your motherboard has a decent number of SATA ports, consider bypassing FreeNAS + HBA and just running some flavor of Windows 10 with Stablebit DrivePool. For example, if you’re a student you might get Windows Server 2016 or 2019 for free through Azure for Students.

Alzabo
Oct 23, 2002

You watched it, you can't unwatch it.


Paul MaudDib posted:

  • If you want to go low-power with a validated ECC platform, go with a 7100 and a C236 board. Or wait for 9300 to really hit the market and go with a C246 board. SuperMicro says 9-series BIOS support will be coming on C246 "later this year" (although that's vague)

I think the X11SSH-F board is going to suit my needs for a new FreeNAS build, thanks for the suggestions!

The price difference between the i3-7100 and a E3-1225 V6 is small enough that I might just spring for the upgrade.

Rexxed
May 1, 2010

Dis is amazing!
I gotta try dis!



Unsurprisingly there's another 8TB Easystore sale on at Best Buy for $130:
https://www.bestbuy.com/site/wd-eas...p?skuId=5792401

manero
Jan 30, 2006



Rexxed posted:

Unsurprisingly there's another 8TB Easystore sale on at Best Buy for $130:
https://www.bestbuy.com/site/wd-eas...p?skuId=5792401

Been waiting for this, and I grabbed two. Thanks!

Edit: I ended up with two EMAZ drives.

manero fucked around with this message at 00:28 on May 27, 2019

Dicty Bojangles
Apr 14, 2001



Illegal Hen

I grabbed two also; time to fill out all the bays in my DS1817+

Enos Cabell
Nov 3, 2004



Snagged the last 3 at my store, replacing two 4 tbs and a 2tb.

Crunchy Black
Oct 24, 2017

CASTOR: Uh, it was all fine and you don't remember?
VINDMAN: No, it was bad and I do remember.




RIP power bill.

Enos Cabell
Nov 3, 2004



Any Unraid users swapped out to a larger drive in the array before? Not sure which is less disruptive, removing an old drive and letting the array rebuild data on the new drive, or copying data off the old drive first before putting in the new drive.

priznat
Jul 7, 2009

Let's get drunk and kiss each other all night.

Why not put in new drive, move folders over from old drive then remove old drive and rebuild?

Or copy folders over maybe to prevent any possible issues

nerox
May 20, 2001


I think there is an app called unbalance that you can use to tell unraid to clear out a disk, then you can remove it. It just shuffles files onto other drives.

Heners_UK
Jun 1, 2002


I've seen a few stories now of people swapping old, smaller drives, for large ones. Call me cheap (lots of people do), but if you have these drives in a parity/raid protected array (and you have backups of truly important stuff that are not raid/parity, which you should, because raid/parity is not backup) then why not simply use the drives until they die? The data is merely a rebuild away at worst.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply
«608 »