I"m using an ASUS M2NPV-VM.  I have 2 PATA drives attached via a promise PATA TX2.

 

I have a telnet window with a tail -f /var/log/syslog running.  Then I start copying about 200mb of files to drive1.. It's all going ok until I click "Refresh" on the main tab in the web interface.. When I do that I get these messages in the syslog:

 

 

Feb 24 14:14:26 Tower kernel: hde: drive_cmd: status=0x51 { DriveReady SeekComplete Error }

Feb 24 14:14:26 Tower kernel: hde: drive_cmd: error=0x04 { DriveStatusError }

Feb 24 14:14:26 Tower emhttp[1287]: smart ioctl: Input/output error

Feb 24 14:14:26 Tower kernel: hdf: drive_cmd: status=0x51 { DriveReady SeekComplete Error }

Feb 24 14:14:26 Tower kernel: hdf: drive_cmd: error=0x04 { DriveStatusError }

Feb 24 14:14:26 Tower emhttp[1288]: smart ioctl: Input/output error

 

 

 

HDE is my parity drive and HDF is my data drive "Drive1".

 

The copy then goes to 0 B/s...

 

after that it will continue going again but very very slow.  like 100-200 KB/s rather than 8MB/s before clicking the refresh button.

 

 

Thanks for the help!

 

I sometimes get these messages too:

 

Feb 24 14:20:39 Tower smbd[1285]: [2007/02/24 14:20:39, 0] smbd/service.c:make_connection(800)

Feb 24 14:20:39 Tower smbd[1285]:  192.168.0.60 (192.168.0.60) couldn't find service .svn

Feb 24 14:23:02 Tower smbd[1285]: [2007/02/24 14:23:02, 0] smbd/service.c:make_connection(800)

Feb 24 14:23:02 Tower smbd[1285]:  192.168.0.60 (192.168.0.60) couldn't find service .svn

Feb 24 14:23:03 Tower smbd[1285]: [2007/02/24 14:23:03, 0] smbd/service.c:make_connection(800)

Feb 24 14:23:03 Tower smbd[1285]:  192.168.0.60 (192.168.0.60) couldn't find service .svn

Feb 24 14:23:03 Tower smbd[1285]: [2007/02/24 14:23:03, 0] smbd/service.c:make_connection(800)

Feb 24 14:23:03 Tower smbd[1285]:  192.168.0.60 (192.168.0.60) couldn't find service .svn

 

 

btw on the main page of the web interface, I have never had an error on any of the drives.

Well I did a little more testing.. I"m not sure it has to do with clicking refresh in the main tab.  I think it's just all the time.

 

I was testing by copying a single 6GB file.  When first starting out the speed bounces around from 30MB/s to 50MB/s and then drops to 0.  Then it bounces back up to 30 for a few seconds and back to 0 for around 10 seconds.

 

 

Here's a screenshot of network monitor showing this behavoir:

 

 

 

Is it my mobo?  The built in network card?  The drives?

 

 

Thanks

So limetech, any ideas for me?  I'd really like to make this product work for me.  I have 8 PATA drives and unraid sounds like the perfect solution... I bought a mobo / cpu and am a little frustrated by this.

 

Today I replaced both the parity and data drive with 2 completely new drives and I'm getting the same exact behavoir.. 30+ MB/s transfer rate which drops to 0 for ~30 seconds.. then back up to 30+ for about 5 seconds then back to 0.. it does this until the copy completes.. 6gb in about 20 minutes.

 

 

btw, the same computer sustains 30+MB/s transfers when booting from an ubuntu cd.

 

In your first post it appears like those drives do not support SMART - are they old drives?

 

Regarding the network utilization graph - what you're seeing is result of network being faster than the server.  What happens is the network bursts in data until the server runs out of network buffer memory.  At this point the network stream is paused.  Eventually the data is written to disk, freeing up buffer space and the network stream is restarted.

 

As for "couldn't find service .svn"... I don't know what this is.  Are you using a Windows PC or something else to connect to the unRAID server?

the couldn't find service .svn comes from my subversion source control service.. It's harmless and I've fixed it anyway..

 

I'm pretty sure my drives don't support smart.. I guess the web interface will poll smart info for every drive everytime you hit a page.. Perhaps a setting that the drive doesn't support smart would be in order?

 

 

As for the network being faster then the HD I guess that makes sense.. But I think this indicates that HD write speed is around 2-5MB/s... can that be right?

 

Reading from these same drives I can pump 30-50MB/s..

 

The drives are older, however they're all ata100 or ata133.

 

 

Thanks for the response!

 

 

So I swapped my gigE switch for a 10/100 switch and ran the test again.

 

 

 

 

the peaks here are 10.5MB/s the valleys are 0MB/s

In your first post it appears like those drives do not support SMART - are they old drives?

 

Regarding the network utilization graph - what you're seeing is result of network being faster than the server.  What happens is the network bursts in data until the server runs out of network buffer memory.  At this point the network stream is paused.  Eventually the data is written to disk, freeing up buffer space and the network stream is restarted.

 

As for "couldn't find service .svn"... I don't know what this is.  Are you using a Windows PC or something else to connect to the unRAID server?

 

Is there a way to increase network buffer memory? My unRaid server has 1GB which most of it, I imagine, is unused, and I see similar things on my writes.

I'm not sure how the write performance can be so low.. I am trying to copy over a bunch of stuff via a 10mbit line.. The spaces between writes get to be so large that the source computer thinks it has timed out..

 

I really want this to work but i'm afraid this level of write performance is going to be unacceptable.

So i tried copying a single 6gb file from 1 drive to another..  _VERY SLOW_.. i was running top during the copy... the unraidd process peaked at about 0.7% cpu utilization... the cp process would sit at 0 for a long time and then briefly spike up to 6%..  I can't watch the network utilization, but I believe the exact same thing is happening .. Why doesn't unraid let more data into the buffer as data is written out?  I dont' think I've ever seen a buffer stop accepting data when full until it's empty..