Difference in write speeds


Recommended Posts

Sorry I forgot to mention that I am not using a cache drive or parity drive.

Write speed to inner cylinders of disks is frequently half of that to the outer cylinders where data density is greater.

 

Also, initial writes are being buffered by system memory.  They will look higher than normal until the memory fills and may skew initial statistics. A reboot would clear the in-memory cache.

 

Other than that, since you did not attach a system log, there is no way to look if there were any other possible reasons for the slowdown.

 

Joe L.

 

Link to comment

Yes. I am transferring data to the server first. Then I will activate the hitachi. I have noticed some annoying things though:

 

1. I have mapped the unraid via windows 7. When i try to get into the mapped drive it takes some time for the contents to show up. When I transfer things to the unraid, windows 7 does a discovery and then it stops and says make sure the unraid is connected. I try the transfer again and this time it goes through.

 

2. Films on my popcorn starting stuttering When I transfer files from the popcorn hour or when I transfer files from my other server to the unraid.

 

Just had a thought. I transferring DVDs from the popcorn to the unraid and I am watching blu rays from the unraid to the popcorn. The popcorn  has only a 100mbps port. Maybe I am reaching the limits of the popcorn?

Link to comment
1. I have mapped the unraid via windows 7. When i try to get into the mapped drive it takes some time for the contents to show up. When I transfer things to the unraid, windows 7 does a discovery and then it stops and says make sure the unraid is connected. I try the transfer again and this time it goes through.

Are the HDDs already spun up? I know with mine sometimes windows XP waits while the drives get spun up. I think (and I could be wrong) but if you are reading a share that spans multiple disks windows will spin up all disks in that share. You can try using cache dirs script and that should help.

 

2. Films on my popcorn starting stuttering When I transfer files from the popcorn hour or when I transfer files from my other server to the unraid.

 

Just had a thought. I transferring DVDs from the popcorn to the unraid and I am watching blu rays from the unraid to the popcorn. The popcorn  has only a 100mbps port. Maybe I am reaching the limits of the popcorn?

 

100mps port on the popcorn hour shouldn't be the issue. Which popcorn hour and is it the latest firmware?

I run C200 (They have the gigabit port but I run mine at 100mbs as the gigabit has been known to freeze stuff) through nfs streaming from unRAID blu ray isos and no problems.

 

What exactly are you running when it is stuttering? When you say transfer are you copying files to unRAID whilst trying to access to read as well?

 

Josh

Link to comment

I have two massive shares using the entire 7 2tb discs. Before I did everything I spun up all of the discs but still the same slow discovery. But i reset my midem router which may help but I can test it as I am doing a transfer of files from the popcorn a200 to the unraid. Currently it's writing at 7 MB/s. Now if I try and watch a blu ray iso from my unraid at the same time, I think this will reach the of the popcorns Ethernet port. Blu rays can run at 5+ MB/s. I am using the popcorns latest firmware.

 

I am using v5 beta 2 unraid.

 

I tried to set the spin down to 15 mins but the GUI has now crashed but oddly the popcorn transfer is still working. What's the best way to get out of this as I am going to let the transfer countinue?

Link to comment

I click on spin up and the GUI froze.

That means nothing.  It indicates your browser was waiting for a response.  The failure to get a response could be almost anything.
 I have plenty of ram!
Do you.   Is it free?  If all used to log errors, you have no FREE memory and processes will be terminated in an attempt to free some.  Since you seem to know you have plenty, I'll let you discover the command to determine the amount of "free" memory.   ;)

 

When I plug the server via hdmi into my amp I can see that the system has crashed whilst in spin down. I cant type in anything. The possible way out is to do control alt delete.

No, that is not the only way.  It might be the only way you are aware of, but you can use "telnet" to log into the server and regain control.   If control-alt-delete works then the array has not crashed.  If it has truly crashed you'll need to use the power button to shut it down.

 

How many disks are you trying to spin up simultaneously?  What specific power supply are you using?  It is capable of the initial start-up surge on the one rail powering all your disks?

 

I'm getting ready to go out to dinner now, but if you still need help later on this evening and you've not regained control of your server by then, post again.    If nobody else has helped, I might be able to figure out what has happened.  

 

In the interim, you might look in the wiki under troubleshooting for how to capture a system log for analysis, as almost everybody willing to help will need to look at the system log for clues.

 

Joe L.

Link to comment

Hi Joe,

 

I tried open the GUI today but it refused to load. I have 4gb of ram in my server.

I am using a Corsair 650 watt PSU with 8 drives. In the drives settings I was trying to apply a 15 minute spin down. When I clicked apply the GUI locked up.

 

As I don't have a parity drive I know I won't be protected from one hdd loss. Does a parity drive check the files that are being transferred to the unraid? If I have to power down by holding the power button (whilst NO data is been written), will this cause any data corruption?

 

When I open telnet what should I type in?

 

 

Link to comment

Hi Joe,

 

I tried open the GUI today but it refused to load.

Can you log in via telnet? or on the system console?
I have 4gb of ram in my server.
It is easy to fill 4Gig if errors are occurring and the system log uses up all available ram.  I asked about "free" memory.  the output of the "free" command will tell more about what is happening.

I am using a Corsair 650 watt PSU with 8 drives.

Is it a single rail supply?  If not, 8 disks on a single 18Amp tail shared with the motherboard and fans could easily be too many when all are asked to "spin up" at the same time.
In the drives settings I was trying to apply a 15 minute spin down. When I clicked apply the GUI locked up.
as I said... takes a lot of current to spin all the disks up at the same time.

As I don't have a parity drive I know I won't be protected from one hdd loss.

Then, at the least,use TerraCopy to migrate your files so it does checksums to ensure the files are transferred without errors
Does a parity drive check the files that are being transferred to the unraid?
Yes it would, as long as you do NOT use a cache drive.  Files written to the cache drive are not protected by parity until the "mover" moves them to the protected array (typically in the middle of the night)
If I have to power down by holding the power button (whilst do data is been written), will this cause any data corruption?
It will result in a partial file on the files currently being written. (or not yet written to the disk and still in the disk buffer cache)

 

When I open telnet what should I type in?

See the prior post.  Look in the wiki under troubleshooting for how to capture the syslog.  Zip it if huge. It will compress well.

If too large for attaching to a post, use pastebin.

 

Joe L.

Link to comment

This is FREE command

 

root@Tower:~# free

                  total        used        free    shared    buffers    cached

Mem:      3887932    3629860    258072          0    321772    3178192

-/+ buffers/cache:    129896    3758036

Swap:            0          0          0

root@Tower:~#

Link to comment

Yes...

this kernel oops... (which I think was fixed in the 4.6 final release)

Dec 23 20:07:19 Tower kernel: BUG: unable to handle kernel paging request at f85

58000

Dec 23 20:07:19 Tower kernel: IP: [<f854e8e4>] md_cmd_proc_read+0x41/0x54 [md_mo

d]

Dec 23 20:07:19 Tower kernel: *pdpt = 0000000001447001 *pde = 000000003761f067 *

pte = 0000000000000000

Dec 23 20:07:19 Tower kernel: Oops: 0000 [#1] SMP

Dec 23 20:07:19 Tower kernel: last sysfs file: /sys/devices/pci0000:00/0000:00:0

3.0/0000:03:00.0/host2/port-2:2/end_device-2:2/target2:0:2/2:0:2:0/block/sdd/sta

t

Dec 23 20:07:19 Tower kernel: Modules linked in: md_mod xor r8169 pata_jmicron j

micron ahci mvsas libsas scst scsi_transport_sas

Dec 23 20:07:19 Tower kernel:

Dec 23 20:07:19 Tower kernel: Pid: 1637, comm: emhttp Tainted: G        W  (2.6.

32.9-unRAID #5) System Product Name

Dec 23 20:07:19 Tower kernel: EIP: 0060:[<f854e8e4>] EFLAGS: 00210212 CPU: 0

Dec 23 20:07:19 Tower kernel: EIP is at md_cmd_proc_read+0x41/0x54 [md_mod]

Dec 23 20:07:19 Tower kernel: EAX: c34f3f38 EBX: fffff41c ECX: 3fffefd6 EDX: fff

ff41c

Dec 23 20:07:19 Tower kernel: ESI: f8558000 EDI: ca73b4c4 EBP: c34f3f04 ESP: c34

f3ef4

Dec 23 20:07:19 Tower kernel:  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068

Dec 23 20:07:19 Tower kernel: Process emhttp (pid: 1637, ti=c34f2000 task=f76b62

60 task.ti=c34f2000)

Dec 23 20:07:19 Tower kernel: Stack:

Dec 23 20:07:19 Tower kernel:  c34f3f38 f854e8a3 00000000 ca738000 c34f3f4c c109

d5f5 00000400 c34f3f3c

Dec 23 20:07:19 Tower kernel: <0> 00000000 00000400 00000000 00000400 b7793000 c

3a88cc0 00000000 c3a88cc0

Dec 23 20:07:19 Tower kernel: <0> 00000400 ca738000 00000001 c3a88cc0 c109d4ea f

ffffffb c34f3f70 c1099f72

Dec 23 20:07:19 Tower kernel: Call Trace:

Dec 23 20:07:19 Tower kernel:  [<f854e8a3>] ? md_cmd_proc_read+0x0/0x54 [md_mod]

Dec 23 20:07:19 Tower kernel:  [<c109d5f5>] ? proc_file_read+0x10b/0x22d

Dec 23 20:07:19 Tower kernel:  [<c109d4ea>] ? proc_file_read+0x0/0x22d

Dec 23 20:07:19 Tower kernel:  [<c1099f72>] ? proc_reg_read+0x56/0x6a

Dec 23 20:07:19 Tower kernel:  [<c1099f1c>] ? proc_reg_read+0x0/0x6a

Dec 23 20:07:19 Tower kernel:  [<c106cbd8>] ? vfs_read+0x8a/0x114

Dec 23 20:07:19 Tower kernel:  [<c106cf6f>] ? sys_read+0x3b/0x60

Dec 23 20:07:19 Tower kernel:  [<c1002935>] ? syscall_call+0x7/0xb

Dec 23 20:07:19 Tower kernel: Code: 55 f0 e8 6a 82 be c8 8d 50 01 29 f2 39 d3 7c

0b 8b 45 0c 89 d3 c7 00 01 00 00 00 8b 45 f0 89 d9 81 c6 3c 3f 55 f8 c1 e9 02 8

9 38 <f3> a5 89 d9 83 e1 03 74 02 f3 a4 5a 89 d8 5b 5e 5f 5d c3 55 89

Dec 23 20:07:19 Tower kernel: EIP: [<f854e8e4>] md_cmd_proc_read+0x41/0x54 [md_m

od] SS:ESP 0068:c34f3ef4

Dec 23 20:07:19 Tower kernel: CR2: 00000000f8558000

Dec 23 20:07:19 Tower kernel: ---[ end trace 20b32b788c63ce70 ]---

Dec 23 20:07:22 Tower kernel: mdcmd (1069): spindown 3

 

Yu can terminate the existing emhttp (in case it is confused) with

killall emhttp

 

Then invoke a new one with

nohup /usr/local/sbin/emhttp &

 

Then see if you can stop the array and reboot.

 

Joe L.

Link to comment

There is a firmware update for the Samsung drives to fix a potential issue with writing to drives and hdparm commands causing corruption and/or bad blocks.

http://sourceforge.net/apps/trac/smartmontools/wiki/SamsungF4EGBadBlocks

 

With those link beat errors I'd turn off autoneg as well

ethtool -s eth0 autoneg off

 

I would run latest 4.6 final codebase on a "production system".

 

 

 

 

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.