I have a 6 disk array that's almost full.  I've recently(last few months) started noticing copies to the array is taking forever, so I timed a some gigabyte file copies to and from the array.  Here are my results:

 

Copy from Array to PC (XP)

disk1 40MB/s

disk2 40MB/s

 

copy to array from PC (XP)

disk1 5MB/s

disk2 2.5MB/s

 

rsync from disk1 to disk2 via telnet using options avz

18MB/S

 

From reading some of the threads, here are some of the culprits people have found:

switch or cable, cleared in my case, 40MB/s from array

Slow data disks, cleared, 18MB/S rsync

slow parity disk, cleared, 18MB/s rsync

 

Here's the output of ethool.

 

root@unRAID:/boot# ethtool eth0

Settings for eth0:

        Supported ports: [ MII ]

        Supported link modes:  10baseT/Half 10baseT/Full

                                100baseT/Half 100baseT/Full

                                1000baseT/Full

        Supports auto-negotiation: Yes

        Advertised link modes:  10baseT/Half 10baseT/Full

                                100baseT/Half 100baseT/Full

                                1000baseT/Full

        Advertised auto-negotiation: Yes

        Speed: 1000Mb/s

        Duplex: Full

        Port: MII

        PHYAD: 1

        Transceiver: external

        Auto-negotiation: on

        Supports Wake-on: g

        Wake-on: d

        Link detected: yes

 

 

 

any ideas?

syslog.txt

Upgrade to 4.7. First make sure that you don't have HPA on any of your disks.

Upgrade to 4.7. First make sure that you don't have HPA on any of your disks.

 

Thank you for the quick reply dgaschk.  Is 4.7 a definite fix or is it just upgrade-to-latest-see-if-it-fixes it?  I searched around, it seems like HPA is isolated to Gigabyte motherboards.  I have an Asus, can I assume I'm free and clear?  I also looked in the syslog, I didn't see any HPA mentioned as per the wiki.

It might fix the problem. HPA has been limited to Gigabyte MBs. As long as none of your drives have ever been in a Gigabyte MB then your good to go. Upgrading may not fix the problem but there are more people willing to help debug the latest version.

Try reading the SMART values:

 

Smartctl -a /dev/sdX, where X is your drive. It could be a failing disk where reads are ok but it is having trouble writing.

 

Have you tried using hdparm -tT /dev/sdX as well to test the disk throughput?

 

Do you have any other disk in your array that you can compare with?

 

And I assumed you've rebooted the array with same results?

 

Any errors in your syslog?

Upgrading may not fix the problem but there are more people willing to help debug the latest version.

 

That's a good point, I'll upgrade if I can't figure it out with some simple troubleshooting.

Try reading the SMART values:

 

Smartctl -a /dev/sdX, where X is your drive. It could be a failing disk where reads are ok but it is having trouble writing.

 

Have you tried using hdparm -tT /dev/sdX as well to test the disk throughput?

 

Do you have any other disk in your array that you can compare with?

 

And I assumed you've rebooted the array with same results?

 

Any errors in your syslog?

 

Here are the results of my hdparm, disk1, disk2, then parity disk in that order, they seem to be fine:

 

root@unRAID:/boot# hdparm -tT /dev/sdg

 

/dev/sdg:

Timing cached reads:  2166 MB in  2.00 seconds = 1084.03 MB/sec

Timing buffered disk reads:  314 MB in  3.01 seconds = 104.32 MB/sec

root@unRAID:/boot# hdparm -tT /dev/sdb

 

/dev/sdb:

Timing cached reads:  2192 MB in  2.00 seconds = 1097.06 MB/sec

Timing buffered disk reads:  422 MB in  3.00 seconds = 140.65 MB/sec

root@unRAID:/boot# hdparm -tT /dev/sdc

 

/dev/sdc:

Timing cached reads:  2124 MB in  2.00 seconds = 1062.65 MB/sec

Timing buffered disk reads:  350 MB in  3.00 seconds = 116.49 MB/sec

 

Other 3 disks have very similar values, nothing out of the ordinary.

 

I don't think there are errors in smarts and syslog.  I attached the smarts with this post.

smarts.txt

copy to array from PC (XP)

disk1 5MB/s

disk2 2.5MB/s

 

rsync from disk1 to disk2 via telnet using options avz

18MB/S

 

This looks like a network problem. Internal transfers to the protected array are at a fairly normal speed. External transfers are very slow. What is the network speed at the PC?

copy to array from PC (XP)

disk1 5MB/s

disk2 2.5MB/s

 

rsync from disk1 to disk2 via telnet using options avz

18MB/S

 

This looks like a network problem. Internal transfers to the protected array are at a fairly normal speed. External transfers are very slow. What is the network speed at the PC?

 

But there's also this:

 

Copy from Array to PC (XP)

disk1 40MB/s

disk2 40MB/s

 

That's I ruled out network being the issue.  If network is slow, it should be slow both ways, not just 1 way. PC's network speed is gigabit.

Key word is "disks almost full"

We do not know how this almost full translate - 99% or 99.9% or....99.99999...% but I believe this is a know fact that you should keep a small amount of free place on any single HD.

And if the disks approach this stage the writing speed will decrease substantially and the copy may even fail...

 

Joe L. probably can update you on this

 

Key word is "disks almost full"

We do not know how this almost full translate - 99% or 99.9% or....99.99999...% but I believe this is a know fact that you should keep a small amount of free place on any single HD.

And if the disks approach this stage the writing speed will decrease substantially and the copy may even fail...

 

Joe L. probably can update you on this

 

 

disk1 is around 88% used, disk2 is 75% used.  These are 500GB and 1TB drives, so that's 64GB and 250GB free.  I think that constitutes a small amount of free space.  I usually see slow down on a drive that is almost full, but I'm experiencing 1/6 the speed here.  Another point is that the internal rsync is fast.  Which means the typical slow down of a drive being almost full probably isn't the factor.

Are all of these test using disk shares? Any user shares?

Are all of these test using disk shares? Any user shares?

 

they were done using disk share, the drives were mapped in xp.  I do not have user share enabled. Except the rsync test of course,  that was run via telnet.

 

 

my 2 cents, have you tried new network cable, tried different port on the gig switch, rebooted the gig switch, rebooted the server, different network card if possible.. just all the usual troubleshooting steps to make sure the hardware is sound..

Are all of these test using disk shares? Any user shares?

 

they were done using disk share, the drives were mapped in xp.  I do not have user share enabled. Except the rsync test of course,  that was run via telnet.

This confuses me. I think I understand disks "were mapped in xp." You mapped "/mnt/diskX" correct?

 

I do not have user share enabled. Except the rsync test of course,  that was run via telnet.

What does telnet have to do with user shares verses disk shares?

 

 

But there's also this:

 

Copy from Array to PC (XP)

disk1 40MB/s

disk2 40MB/s

 

That's I ruled out network being the issue.  If network is slow, it should be slow both ways, not just 1 way. PC's network speed is gigabit.

 

You would thing that would be the case, but it often is not.  I'm not saying you are wrong, but it is still something that needs to be checked.

 

Can we get a full breakdown of your hardware? motherboard, CPU, GB of RAM, etc please.

my 2 cents, have you tried new network cable, tried different port on the gig switch, rebooted the gig switch, rebooted the server, different network card if possible.. just all the usual troubleshooting steps to make sure the hardware is sound..

 

I suspected the network too, but as you can see, my copy from the unraid server is at 40MB/s

Are all of these test using disk shares? Any user shares?

 

they were done using disk share, the drives were mapped in xp.  I do not have user share enabled. Except the rsync test of course,  that was run via telnet.

This confuses me. I think I understand disks "were mapped in xp." You mapped "/mnt/diskX" correct?

 

Correct.

 

I do not have user share enabled. Except the rsync test of course,  that was run via telnet.

What does telnet have to do with user shares verses disk shares?

 

I was just saying all the tests were run via the disk shares, except for the rsync test, which was run via telnet.  Meaning if the slow down was caused by disk share, the rsync would not be affected.

But there's also this:

 

Copy from Array to PC (XP)

disk1 40MB/s

disk2 40MB/s

 

That's I ruled out network being the issue.  If network is slow, it should be slow both ways, not just 1 way. PC's network speed is gigabit.

 

You would thing that would be the case, but it often is not.  I'm not saying you are wrong, but it is still something that needs to be checked.

 

Can we get a full breakdown of your hardware? motherboard, CPU, GB of RAM, etc please.

 

I'll get these info when I get home tonight, I don't have the details, but it's an approved ASUS board.  AMD chip with 1 GB of ram.

It is probably good idea to upgrade to 4.6 if not 4.7 anyway. I do recall seeing some posts about major speed updates but do not remember when this happened - Joe L. or the other long timers can provide an update on this.

 

This is from the history:

=====================================

Changes from 4.5-beta3 to 4.5-beta4:

"Disable NCQ on all disk devices that support NCQ.  This typically results in much better write throughput"

 

Changes from 4.5-beta7 to 4.5-beta8:

Increased write peformance and added three user-tunables:

======================================

The kernel has updated too...

 

Second - you have an nVidia based motherboard (with nVidia SATA ports and nVidia "reverse engineered" LAN driver). Cannot comment more on that.

my 2 cents, have you tried new network cable, tried different port on the gig switch, rebooted the gig switch, rebooted the server, different network card if possible.. just all the usual troubleshooting steps to make sure the hardware is sound..

 

I suspected the network too, but as you can see, my copy from the unraid server is at 40MB/s

 

there are different transmit and receive pairs on the cabling so just maybe there is a hardware error somewhere's along the data line....

just putting that out there (stranger things have happened)

So I have the ASUS M2NPV-VM with 1GB of ram and AMD processor.  Switch is an ASUS GX-d1081 gigabit switch I recently purchased after the trendnet 5port gigabit stopped working.  So the switch is new. 

It is probably good idea to upgrade to 4.6 if not 4.7 anyway. I do recall seeing some posts about major speed updates but do not remember when this happened - Joe L. or the other long timers can provide an update on this.

 

This is from the history:

=====================================

Changes from 4.5-beta3 to 4.5-beta4:

"Disable NCQ on all disk devices that support NCQ.  This typically results in much better write throughput"

 

Changes from 4.5-beta7 to 4.5-beta8:

Increased write peformance and added three user-tunables:

======================================

The kernel has updated too...

 

Second - you have an nVidia based motherboard (with nVidia SATA ports and nVidia "reverse engineered" LAN driver). Cannot comment more on that.

 

I'll definitely upgrade after some simple checks and troubleshooting doesn't work out.

my 2 cents, have you tried new network cable, tried different port on the gig switch, rebooted the gig switch, rebooted the server, different network card if possible.. just all the usual troubleshooting steps to make sure the hardware is sound..

 

I suspected the network too, but as you can see, my copy from the unraid server is at 40MB/s

 

there are different transmit and receive pairs on the cabling so just maybe there is a hardware error somewhere's along the data line....

just putting that out there (stranger things have happened)

 

That was my immediate thought.  Just because network transfers are good in one direction doesn't mean that all must be okay in the reverse direction.

 

Try changing cables (or even simply reverse the cables), and move to different switch ports.  Of course, it could still be a hardware fault on the network interface on one or other of your machines.  Do you have another network device that you can time transfers with ... in both directions - to/from unRAID and to/from the original client computer?

my 2 cents, have you tried new network cable, tried different port on the gig switch, rebooted the gig switch, rebooted the server, different network card if possible.. just all the usual troubleshooting steps to make sure the hardware is sound..

 

I suspected the network too, but as you can see, my copy from the unraid server is at 40MB/s

 

there are different transmit and receive pairs on the cabling so just maybe there is a hardware error somewhere's along the data line....

just putting that out there (stranger things have happened)

 

That was my immediate thought.  Just because network transfers are good in one direction doesn't mean that all must be okay in the reverse direction.

 

Try changing cables (or even simply reverse the cables), and move to different switch ports.  Of course, it could still be a hardware fault on the network interface on one or other of your machines.  Do you have another network device that you can time transfers with ... in both directions - to/from unRAID and to/from the original client computer?

 

I really hope it's as simple as that.  I'm running a few tests now with a different cable and port.  I'll also test some transfers via my laptop.  Results to come.