Login   Register
 Login    Register
Server Nerds Thread
Neil T
User avatar
Dead chuffed
 
Rank: Moderator
Posts: 47,439 | Offline
Joined: Apr 9, 2008
Der Lobo wrote:
Neil T wrote:
Der Lobo wrote:I'm doing backblaze for all the non video/audio data. Highly recommended, 5 bucks a month, offline backup for everything that I really don't want to lose (contracts, insurance, photos etc). Peace of mind for me.

I hear they have a speed cap when uploading? Which is why I haven't gone with them yet. How do you find the upload speed?

Oh, and the fact they don't support NAS or Lunix. So, I'd have to sync everything to a PeeCee all the time?

Don't know about speed cap. Last uploads were 5 mbits, pretty close to my line limit.

Yeah, no Linux for the unlimited. Only PC and mac clients. Works for me, windows as main desktop, syncing to unraid and backing up to backblaze. Got roughly 1tb backed up.

Hmmm. Thanks for the info. I may look in to this. I have a few spare 2TB drives lying around, so could put them in my workstation and sync from unRAID to those and then to BackBlaze.
Or, even better, assign an entire drive on my server to the Windows VM running on it and run Backblaze in the VM. It could take a while, even at my line speed to upload, so better to have the data on a machine that's always on.

Image

Der Lobo
User avatar
Yall loosers
 
Rank: Vetted
Posts: 15,270 | Offline
Joined: Apr 15, 2013

Neil T wrote:[8 TEEB DRIVE]

Where did you buy from and how much?

Let us know (mebbe in the Server thread) how you get on with the write speeds and reliability. It seems SG has improved, but I've been burned so many times with dead drives from them.


http://www.bhphotovideo.com/c/product/1107004-REG/seagate_st8000as0002_archive_hdd_8tb_sata.html

It was 216 (including tax/shipping) when I ordered. 8TB prices haven't moved much in the last few months, it looks like.

preclear is running. last time it took 39h, let's see when this one finishes.


Just ran a speedtest against a one disk share (WDC_WD60EFRX share disk, plus ST8000AS0002-1NA parity disk) with a couple of VMs active to avoid memory cashing:

"Date","Time","Computer","IP Address","MAC Address","Folder Or Server IP","Packet Size","# Packets","# Streams","RTT (LST Server)","Write Speed","Units","Read Speed","Units","Write Time","Read Time","Status","Notes"
"2016-03-05","11:48:09","PC","192.168.0.99","XC-XX-XC-XX-XX-XX","\\TOWER\test speed","1,000,000,000","10","","","50.3411210","MBps","94.3194930","MBps","198.6447619","106.0226224"," (Average)",""

50MB avg for 10GB written. I'll do the same once the 8tb is clear and added to the pool.

Neil T
User avatar
Dead chuffed
 
Rank: Moderator
Posts: 47,439 | Offline
Joined: Apr 9, 2008

50 is pretty good. Can't remember how big the cache tracks are on the drive, so mebbe more needs to be written to exhaust the cache.

Der Lobo
User avatar
Yall loosers
 
Rank: Vetted
Posts: 15,270 | Offline
Joined: Apr 15, 2013

Neil T wrote:50 is pretty good. Can't remember how big the cache tracks are on the drive, so mebbe more needs to be written to exhaust the cache.


looks like you were right:

"Date","Time","Computer","IP Address","MAC Address","Folder Or Server IP","Packet Size","# Packets","# Streams","RTT (LST Server)","Write Speed","Units","Read Speed","Units","Write Time","Read Time","Status","Notes"
"2016-03-05","13:28:41","PC","192.168.0.99","90-XX-34-XX-E4-XX","\\TOWER\test speed","5,000,000,000","10","","","30.6633300","MBps","88.0447220","MBps","1,630.6122184","567.8932138"," (Average)",""

50GB written, avg 30, max 62, min 21. I'll have to graph it for a conclusive result but it looks like after 8GB it went down to low 30s. at least some data to compare to the 8TB

Neil T
User avatar
Dead chuffed
 
Rank: Moderator
Posts: 47,439 | Offline
Joined: Apr 9, 2008

I ended up moving data (or trashing it) from the failed disk, as I was concerned about having no redundancy. In theory, I should have just bought another 4TB drive, but I have no more room for extra drives once the replacement arrives.

Moving data from a simulated drive is slow! Especially when the drives you're moving to are also doing the simulation. I was getting about 22MB/sec.

But that's all done now and I rebuilt parity, so it's all protected now. That also lets me do a proper preclear when the replacement arrives.

Der Lobo
User avatar
Yall loosers
 
Rank: Vetted
Posts: 15,270 | Offline
Joined: Apr 15, 2013

That makes sense. Yeah reading files that are not on a single disk but need to be xor'd from all disks creates a lot of IO. Glad that it worked without a problem. Do you do that locally on the box via mc or remotely? Don't think network speed would matter at that point.

With 40h preclear times I really don't want to add a 'raw' drive to the array. Thankfully I have one more empty slot (and one 6tb warm standby disk) that I can use. 16 more teebs, that'll last a while.

Neil T
User avatar
Dead chuffed
 
Rank: Moderator
Posts: 47,439 | Offline
Joined: Apr 9, 2008

Mostly it was screen and Midnight Commander.

I did move some data to the 2 Teeb drives in my other box.

Der Lobo
User avatar
Yall loosers
 
Rank: Vetted
Posts: 15,270 | Offline
Joined: Apr 15, 2013

Had a huge power outage this morning at 6. I really hope the preclear finished before that ... power was still out when I left for work so I couldn't check. The suspense ...

Neil T
User avatar
Dead chuffed
 
Rank: Moderator
Posts: 47,439 | Offline
Joined: Apr 9, 2008

How many passes are you doing?

Der Lobo
User avatar
Yall loosers
 
Rank: Vetted
Posts: 15,270 | Offline
Joined: Apr 15, 2013

Neil T wrote:How many passes are you doing?

Just did a single one. For a data drive I'm fine with just one.

Ps: looks like it finished but for some reason I lost the log files during the reboot. Might just run another pass.

Neil T
User avatar
Dead chuffed
 
Rank: Moderator
Posts: 47,439 | Offline
Joined: Apr 9, 2008

There are posts on the LT forum saying if you do 3 passes, that equates to an entire year's duty cycle!
Because: Seagate and their low duty cycles on consumer drives.

Dont Tell Ray
User avatar
world's greatest no-goodnik
 
Rank: Vetted
Posts: 28,477 | Online
Joined: Jan 30, 2008

what does a duty cycle mean on hdd? and how is one mfg diff from other?

Neil T
User avatar
Dead chuffed
 
Rank: Moderator
Posts: 47,439 | Offline
Joined: Apr 9, 2008

How many Teebs written (although forum peeps think it means written or read) per year that the mfr rates it under warranty.

Only Seagate seem to use it and it's quite low. However, if you return a drive under warranty, no-one has seen them refuse the RMA. They just replace the drive before looking at it (as Hitachi just did with mine).

Der Lobo
User avatar
Yall loosers
 
Rank: Vetted
Posts: 15,270 | Offline
Joined: Apr 15, 2013

Der Lobo wrote:Just ran a speedtest against a one disk share (WDC_WD60EFRX share disk, plus ST8000AS0002-1NA parity disk) with a couple of VMs active to avoid memory cashing:

Code: Select all
"Date","Time","Computer","IP Address","MAC Address","Folder Or Server IP","Packet Size","# Packets","# Streams","RTT (LST Server)","Write Speed","Units","Read Speed","Units","Write Time","Read Time","Status","Notes"
"2016-03-05","11:48:09","PC","192.168.0.99","90-XX-34-XX-E4-XX","\\TOWER\test speed","1,000,000,000","10","","","50.3411210","MBps","94.3194930","MBps","198.6447619","106.0226224"," (Average)",""

50MB avg for 10GB written. I'll do the same once the 8tb is clear and added to the pool.


Der Lobo wrote:
Code: Select all
"Date","Time","Computer","IP Address","MAC Address","Folder Or Server IP","Packet Size","# Packets","# Streams","RTT (LST Server)","Write Speed","Units","Read Speed","Units","Write Time","Read Time","Status","Notes"
"2016-03-05","13:28:41","PC","192.168.0.99","90-XX-34-XX-E4-XX","\\TOWER\test speed","5,000,000,000","10","","","30.6633300","MBps","88.0447220","MBps","1,630.6122184","567.8932138"," (Average)",""

50GB written, avg 30, max 62, min 21. I'll have to graph it for a conclusive result but it looks like after 8GB it went down to low 30s. at least some data to compare to the 8TB


ST8000AS0002-1NA share disk, plus ST8000AS0002-1NA parity disk

Code: Select all
"Date","Time","Computer","IP Address","MAC Address","Folder Or Server IP","Packet Size","# Packets","# Streams","RTT (LST Server)","Write Speed","Units","Read Speed","Units","Write Time","Read Time","Status","Notes"
"2016-03-10","18:15:40","PC","192.168.0.99","90-XX-34-XX-E4-XX","\\TOWER\test 8tb disk","1,000,000,000","10","","","50.3909870","MBps","97.1346320","MBps","198.448187","102.9498928"," (Average)",""

Code: Select all
"Date","Time","Computer","IP Address","MAC Address","Folder Or Server IP","Packet Size","# Packets","# Streams","RTT (LST Server)","Write Speed","Units","Read Speed","Units","Write Time","Read Time","Status","Notes"
"2016-03-10","18:59:43","PC","192.168.0.99","90-XX-34-XX-E4-XX","\\TOWER\test 8tb disk","5,000,000,000","10","","","29.9869310","MBps","66.1420710","MBps","1,667.393025","755.9485096"," (Average)",""

50avg for 10GB written, 30avg for 50GB written. I checked the config twice to make sure I didnt use the same 6TB data disk. Nearly identical. Maybe the 8TB parity disk is the limiting factor. Anyway, good enough for me.

Neil T
User avatar
Dead chuffed
 
Rank: Moderator
Posts: 47,439 | Offline
Joined: Apr 9, 2008

You could always add a large SSD and use it as a cache disk is supposed to be used. :D
I use mine not in that way, but as an "apps" drive and don't mind about slow write speeds to the array.

Hitachi shipped my replacement drive on Tuesday the 8th and said it'd arrive in 7-10 days. I assumed they were being conservative, but apparently not. It appears to have been strapped to a tortoise.

Not as bad as ordering stuff from China, though. They have a very different idea of "Express Shipping" than we do over here... :shocker:

Who is online

Users browsing this forum: No registered users and 1 guest

cron