My transfer speeds are slow.

[Chris230291]

Hello. I have just upgraded my FreeNas server with the total amount of storage drives it can take, after i learned you can't add AND keep redundancy.
The transfer speeds are now worse than before (though i have no numbers to prove this). My network monitor is registering 1.2MB/s (total) from my PC when i try to send a file from my computer to the server. A 2.17 GB folder is taking 30 mins to transfer! Now i think i may have a bottle neck somewhere in my network, but first can someone help me optimise my FreeNas settings please?

This is my server...

i7 860s
16 GB Corsair vengeance
Gigabyte GA-H55M-UD2H
HighPoint 2720SGL
8 X 2TB WD Greens

Running off a 2GB USB flash drive.

The server is used as a media/file server. It's mostly ripped movies (BluRay + DVD), TV shows and CD's, although other things are on there too.

Basically i'd like to know what settings i should run to get the best possible transfer speeds i can.
Im sure i've probably missed out a lot of info here but I'm not entirely sure what i need to provide. As such ANY help is really appreciated!

Note: I'm at a stage now where i still have enough storage space elsewhere if i need to format the server and start over.

Many thanks,
Chris.

 

[Stephens]

Hello. I have just upgraded my FreeNas server with the total amount of storage drives it can take, after i learned you can't add AND keep redundancy.

I'm not sure where you got that impression. If you haven't already, read the FreeNAS documentation and cyberjock's ZFS primer.

My network monitor is registering 1.2MB/s (total) from my PC when i try to send a file from my computer to the server. A 2.17 GB folder is taking 30 mins to transfer! Now i think i may have a bottle neck somewhere in my network, but first can someone help me optimise my FreeNas settings please?

It's easy to check for a networking bottleneck using iperf (FreeNAS) and jperf (Windows). Do it.

This is my server...

i7 860s
16 GB Corsair vengeance
Gigabyte GA-H55M-UD2H
HighPoint 2720SGL
8 X 2TB WD Greens

What version of FreeNAS are you running? Your system is plenty powerful. Do you have compression compression? What about dedupe? Show us the output of "zpool status" from a FreeNAS console window (you can use SSH if you have that enabled).

Basically i'd like to know what settings i should run to get the best possible transfer speeds i can.

With a machine that powerful, you shouldn't need to do a lot of tinkering. And by much, I mean any.

Note: I'm at a stage now where i still have enough storage space elsewhere if i need to format the server and start over.

Good, because it's quite possible you'll need to.

 

[cyberjock]

I'm with Stephens on the comment you shouldn't need tweaks. I have a machine that is less powerful than yours in every way and it can saturate 2 Gb LAN ports simultaneously.

First things I'd check:

-You failed to mention what version you are using. If you are using anything except 8.3.0-RELEASE-x64 I'd recommend you upgrade to it.
-Try enabling the auto-tuning feature.
-You made no mention of how you upgraded exactly. I'll assume you replaced 1 disk at a time with your larger 2TB disks. If you did I would do a df and see if the zpool is the new larger size you were expecting.
-When was the last time you did a scrub? If you haven't done one in 30 days I'd do one. If you have a disk failing(which may be the reason for your horrible performance) then you'll see lots of errors when you do zpool status. If you haven't done one aside from the resilvering when upgrading to 2TB drives I'd consider doing one. I always do a scrub after all of the new disks are in and the old disk removed just to see if all is well.
-If you don't have email setup and SMART emails enabled for all drives that support it I'd strongly recommend that you do. In the event of a failing disk you'll get emails that the disk is having problems.

This really sounds like a disk that is failing or a very very misconfigured system. Considering you mentioned upgrading your disks recently I doubt its a misconfigured system.

 

[Chris230291]

Thanks for the reply!

I'm not sure where you got that impression. If you haven't already, read the FreeNAS documentation and cyberjock's ZFS primer.

Hmmm, I made a topic and i was told i needed to start with all of the drives? Else i loose redundancy? :S Nevermind.

It's easy to check for a networking bottleneck using iperf (FreeNAS) and jperf (Windows). Do it.

That sounds exactly like what i need to check my network. Will it run OK inside a virtual machine? I don't have a windows computer.

What version of FreeNAS are you running? Your system is plenty powerful. Do you have compression compression? What about dedupe? Show us the output of "zpool status" from a FreeNAS console window (you can use SSH if you have that enabled).

I am using "FreeNAS-8.3.0-RELEASE-p1-x64 (r12825)"
I dont think i have compression enabled... or dedupe ? Is there any way i can upload all of my settings for someone to check?
Here, Is this what you mean??...

With a machine that powerful, you shouldn't need to do a lot of tinkering. And by much, I mean any.



Good, because it's quite possible you'll need to.

That reassuring, thanks. Like i said, before it was much quicker than it is now. I have obviously done something wrong somewhere along the line.

Thanks again,
Chris.

 

[Chris230291]

Hello Cyberjock. When i say upgrade, what i mean is i backed up the server, started over with more HDD's and a fresh OS, then transferred the files back over. Previously i had 3 x 2TB disks, but now i have 8.

A scrub? I have never done a scrub... ever. My server was running fine before i added more drives, obviously i must have chosen some bad settings in the process :/

The disks are all new apart from the original 3. Im pretty sure it's a misconfigured system. The zpool thingy seems ok?
I like the sound of this email notification system, i will look into that when everything is running sweet.

Thanks again,
Chris.

 

[cyberjock]

Ok, if you haven't done a scrub that's not good. Scrubs verify that your data is good so that if a disk fails your zpool is actually consistent. You need to pull out the manual and read up on scrubs and set that up. I recommend 14 days between scrubs for non-enterprise class drives(read: the ultra-expensive drives). Without doing scrubs there are some things that can go wrong that you will not be aware of until its too late. This is MAJOR oversight and should be fixed. However, it is likely not the problem you are having. You can start a scrub manually by the command zpool scrub yourzpoolname. If you have a failing disk that is responsible for your problems, you'll see errors start counting up like crazy when you do a zpool status. Scrubs are time consuming and do slow the server down, so don't run it if you want peak performance. Of course, with the performance you are getting I doubt you're using the server for much. Hint: The first thing I think of when people have performance that is less than 5MB/sec is that a disk is failing(or failed). A single failing disk can cause a zpool to be so incredibly slow as to be almost useless.

If you haven't setup the emails, you should do that. Again, consult the manual for instructions. The emails, when properly setup, will warn you of disks that are misbehaving and other issues with the server along with nightly and monthly reports. When my friend's server started having bad sectors he started getting emails. We knew exactly which hard drive it was and replaced it without incident.

Can you post the output of zpool status from an SSH session as root and post it inside of [code][/code]

Edit: After posting this I decided that I recommend you read the manual cover to cover. It takes less than 2 hours, and will teach you alot about extra features that can make your life easier and safer as an admin. Keep a piece of scratch paper hand so that if you read about something that sounds like a great idea you write it down. I had zero FreeBSD(or Unix or Linux) experience and just reading the manual gave me enough info to setup my first server with no problems at all. Personally, I keep a paper copy of the manual with my server at all times for reference. I actually use it more often for helping others on the forum than I do myself.
[

 

[Chris230291]

Ok i will look into what exactly a scrub is and how best to set it up once i'm sorted.
Like you said it doesnt appears to be a drive failure as this here says "No known data errors"...

Code:

Chriss-Mac-Pro:~ Chris$ ssh -t  root@10.0.1.200 -p 22 "cd /root && exec \$SHELL"
The authenticity of host '10.0.1.200 (10.0.1.200)' can't be established.
RSA key fingerprint is 75:2c:aa:e8:9a:c2:35:04:b8:ed:0f:d9:3c:63:bb:93.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added '10.0.1.200' (RSA) to the list of known hosts.
root@10.0.1.200's password: 
[root@Server] ~# zpool status
  pool: Raid-Set
 state: ONLINE
  scan: none requested
config:

	NAME                                            STATE     READ WRITE CKSUM
	Raid-Set                                        ONLINE       0     0     0
	  raidz1-0                                      ONLINE       0     0     0
	    gptid/2ab59d51-871a-11e2-a99a-1c6f65a764db  ONLINE       0     0     0
	    gptid/2b616365-871a-11e2-a99a-1c6f65a764db  ONLINE       0     0     0
	    gptid/2c12211e-871a-11e2-a99a-1c6f65a764db  ONLINE       0     0     0
	    gptid/2cb904c3-871a-11e2-a99a-1c6f65a764db  ONLINE       0     0     0
	    gptid/2d596836-871a-11e2-a99a-1c6f65a764db  ONLINE       0     0     0
	    gptid/2df4b9ab-871a-11e2-a99a-1c6f65a764db  ONLINE       0     0     0
	    gptid/2e848c15-871a-11e2-a99a-1c6f65a764db  ONLINE       0     0     0
	    gptid/2f1ec4a4-871a-11e2-a99a-1c6f65a764db  ONLINE       0     0     0

errors: No known data errors
[root@Server] ~# 

- Did i do that right?


I will have to have a read through this manual then. To set up what i had and have now i used YouTube tutorials.

Cheers,
Chris.

 

[cyberjock]

The zeros are a good indicator that the drives aren't bad, but they aren't a "sure thing". Those values also change on every reboot and if you do a zpool clear.

I'd get SMART info for all of the drives to see if anything looks bad and I'd start a scrub unless you plan to do anything that will need the computer for the next 2-24 hours. The scrub is dependent on the speed of your drives, the amount of data in your zpool, and your CPU's ability to do the scrub calculations. I'd expect that your zpool if it were 100% full would take between 3 and 6 hours. Of course, other factors can cause this value to change.

Another easy test(and this is grasping for straws) is to run a RAM test. Normally I'd expect bad RAM to cause server instability, but considering the test is easy to perform(and free) why the heck not? At least you can rule that out as a cause.

 

[Chris230291]

Thanks.
How do i get SMART info for my drives?
I will run a scrub and see what happens. Will i still be able to access files whilst it's... scrubbing?
I should hope the RAM is fine as it's brand new! Having said that, when i added additional drives i also add in the new RAM.

EDIT: I haven't touched anything other than unplugging, then replugging the network cable back into the router and now i reach 10 MB/s. What kind of maximum speeds should i be getting?

Here's a list of my network hardware...

Airport Extreme - Specs
Airport Express - Specs
Pluscom 5 Port Switcher - Specs

Here is a rough diagram of my setup...



Any feedback is much appreciated.

Chris.

 

[Chris230291]

Hello. I have tried setting up a scrub using the Freenas manual but i want to make sure i understand it right.

If i understand right, i have it set to scrub every Sunday at 06:00? I based the 7 day gap on the recommendation for consumer drives and chose a time where an impact on performance won't matter.


Can anyone please confirm this?