Jump to content
Covecube Inc.
  • 0

Slow Write Speeds to the Drivepool


Question

I am having some very sporadic write speeds to the drivepool, it will peak at 16MB/s, go as low as 0MB/s, completely stopping, and finally bouncing between 400KB/s to 1.5MB/s, However, when I write a file from the drivepool to any computer on my network the transfer is fine, between 60MB/s to 90MB/s.  Write speeds to a folder on a non-pool disk is also between between 60MB/s to 90MB/s.

 

I am running StableBit.DrivePool_1.3.3.7563_Release.and the pool is showing 100%. My drivepool is on a frankenbuild WHS 2011.  All of my clients are running Windows 8.  Pinging from my clients to my server is <1ms.  

 

I also tried using Robocopy or Teracopy but neither didn't help.  I also performed the following optimization tasks obtained from wegotserved.com:

 

For the NIC:
Flow control: on (client and server)
"checksum" or "Offload": Off (client and server)
jumbo frames: off or default (client or server)
run "netsh interface tcp set global autotuninglevel=highlyrestricted" in an elevated console (command prompt) on *all* computers (Vista/Win7/Server 2008(R2)/WHS2011/SBS2011e)

Services/programs:
Windows Media Player Network Sharing Service: Disabled/manual 
Windows Search Service: Disabled (and probably a good idea to move it to another drive anyhow, or at least d:\)
Install the "Remote Differential Compression" feature on the server, or uninstall it from all Win7/WHS2011 computers

Misc:
Run chkdsk on all the disks. Full is best, but just to make sure
Defrag.

 

Is there anything that I can do to remedy this?

Link to post
Share on other sites

23 answers to this question

Recommended Posts

  • 0

It looks like you are saying that writing to your pool is slow and reading from the pool is fast.  Do you get the same results when reading and writing using a non-pooled drive on your WHS?  This will help you determine if it is DrivePool or something else.  Your network connection is 2-way communication no matter which way your transfers are going, so I would not think it would be a network issue.

 

If you find that the problem is related to DrivePool (ie your speeds both ways on a non-pooled drive are fast), then you may want to check your SMART data for any errors or warnings on any of your drives.  If you don't have anything installed to read SMART data, then you can install Scanner or HD Tune on a trial basis and check it.  Sometimes a failing disk can cause the system to slow down even if it is not the disk currently being written to.  Is this a new install or was it working fine and then this problem just started happening.  You may also check Event Viewer to see if anything is logged there.

Link to post
Share on other sites
  • 0

Thanks for the replies. Read and write speeds to nonpooled drives are normal, fast from all of my client PCs. It's only slow when I copy or move a file to the pool from any of my client PCs. When copying or moving files from my pool it is just as fast as my read /write speeds that I get when moving files between my client PCs and non pooled drives on my server.

 

Scanner shows that all drives are healthy. How do I do burst test? Personally it seems like all signs are pointing to a problem with my pool or pool setting.

Link to post
Share on other sites
  • 0

To do a Bust Test in Scanner, you just right-click on a disk and you will see the option in the context menu. Once you have tested and verified the read/write speeds of all your drives outside the pool, you can try some simple DrivePool settings changes to determine where the problem might be.  Under your Pool Options, you can make sure that Network I/O boost, as drashna suggested, is turned on under Performance and you can temporarily turn off Real-time duplication and turn off Balancing to see if anything changes.

Link to post
Share on other sites
  • 0

To do a Bust Test in Scanner, you just right-click on a disk and you will see the option in the context menu. Once you have tested and verified the read/write speeds of all your drives outside the pool, you can try some simple DrivePool settings changes to determine where the problem might be.  Under your Pool Options, you can make sure that Network I/O boost, as drashna suggested, is turned on under Performance and you can temporarily turn off Real-time duplication and turn off Balancing to see if anything changes.

Ok thanks, I will try your suggestions when I get home this evening.  Real time duplication was already off but I cannot verify the other settings until I get off.

Link to post
Share on other sites
  • 0

Also, you can use the "Burst test" in StableBit Scanner, to see if there is an issue with the drive/controller communication.

 

Also, have you tested out writes to each drive in the pool?

 

And, do you have Network IO Boost enabled?

Ok, burst test results indicated a average speed of 151MB/s per drive  I don't see a setting to turn on or off Network IO Boost.  I didn't find it under settings, drivepool.  Is this a hidden feature?

Link to post
Share on other sites
  • 0

Could you do this then?

http://wiki.covecube.com/StableBit_DrivePool_Log_Collection

That's a low level kernel logging feature, and may indicate where the issue is.

 

 

Also, if you have Scanner installed, you could run the "burst" test for a few hours or overnight, and that will see if there is an issue with the drive/controller.

Link to post
Share on other sites
  • 0

Could you do this then?

http://wiki.covecube.com/StableBit_DrivePool_Log_Collection

That's a low level kernel logging feature, and may indicate where the issue is.

 

 

Also, if you have Scanner installed, you could run the "burst" test for a few hours or overnight, and that will see if there is an issue with the drive/controller.

Ok, I will run burst test tonight, is there a way to run it on all drives continuously?  I have 12 in total.  Also, how do I tell if it is a problem with the controller oppose to the drive? If the majority of the drives are showing good does that point to the drives that are not showing good as the problem?

Link to post
Share on other sites
  • 0

Ok, I don't think it is a problem with the software. I have 12 HDDs spread out over two sans digital 8 bay cases and oddly enough one side (4 bays) is not working now (the case lights are on but WHS2011 does not detect them nor does the raid card bios during the boot process).

 

I have tested the drives by pulling each one out and running dianostic tools on each drive and all passed. 

 

Next I moved 4 drives to the 4 empty slots and WHS2011 and the raid card bios recognized all 12 drives.

 

IMO, this points to either the sata cable, one of the raid card ports, or one of the ports on the case is bad.  I will get a cable today and rule it out.

 

My questions are how often from personal experience does external sata cabling go bad?  How likely for only one port of a raid card or one port of an enclosure to go bad?

Link to post
Share on other sites
  • 0

Check your power connections make sure all the pins are intact I had a similar problem when 1 of my connections on a lead was faulty

 

I would also power it all down and give it all a good clean sometimes connections can get a little grubby and don't always make a sound connection.

 

It is very rare for leads to become faulty once a sound connection has been made unless they get bent or knocked by accident there cheap enough to replace so its worth trying

Link to post
Share on other sites
  • 0

Check your power connections make sure all the pins are intact I had a similar problem when 1 of my connections on a lead was faulty

 

I would also power it all down and give it all a good clean sometimes connections can get a little grubby and don't always make a sound connection.

 

It is very rare for leads to become faulty once a sound connection has been made unless they get bent or knocked by accident there cheap enough to replace so its worth trying

Thanks for the response.

 

Yea, I checked the one cable but I didn't notice any abnormalities but I am not a Sata cable expert.  Grubby? How can a cable that is inserted get grubby?  It is not like it's outside or near the ocean.

 

My server and cases are in a rack and don't suffer from bumping or moving in general.

Link to post
Share on other sites
  • 0

Dust and stuff... I live in an area that everything gets dusty, so I have to clean out my server and other computers every six months.

 

 

As for burning out cables... There isn't a cable type that I haven't burned out at least once. that includes SATA cables, cat5(e), and even Coax. :P

Link to post
Share on other sites
  • 0

Dust and stuff... I live in an area that everything gets dusty, so I have to clean out my server and other computers every six months.

 

 

As for burning out cables... There isn't a cable type that I haven't burned out at least once. that includes SATA cables, cat5(e), and even Coax. :P

You must live in the midwest or Arizona, lol.  I cannot wait to get home, so do you agree it is not likely one port on either the sata card or 8 bay case?

Link to post
Share on other sites
  • 0

I live in the San Deigo area, so yeah, lots of dust. That, and lots of animals. 

 

I'd be more inclined to believe it is the box it's in or maybe the controller card. That's where I'd put my money.

But swapping out the cable should be trivial, and very easy/quick to test.

Link to post
Share on other sites
  • 0

I've found "positive pressure" filtered cases can noticeably reduce dust and salt intake; you just have to remember to clean the filters every so often to maintain the pressure ratio (and also so that the case doesn't overheat - and for this I can plug Scanner, with it's emailed and spoken overheat warnings).

Link to post
Share on other sites
  • 0

Yeah, messing with stuff late at night is a recipe for disaster. 

And yeah, cable management can be a PITA when you need to move stuff around. That's why I tend to just jury-rig things into place. Because I like to tinker. A lot. :)

 

 

 

As for my server, I think I already have. I'm not sure. But the big thing is the HighPoint RocketRAID 2720SGL card. 2x SAS cables, that break out to 8 HDDs total. :)

Link to post
Share on other sites
  • 0

Ok, I finally got around to finishing my troubleshooting of my setup and it turned out to be my rocketraid 2314 card.:(  Luckily I had a backup Addonics (ADS3GX4R5-E), the bad news is that it has a PCI bus, which I was all out of.  I had to drive a country mile to microcenter to pick up a PCIe1X usb 4 port card, so I could swap some cards around.  My Pool is re duplicating now, around 40% but all looks good so far.

 

Thanks everyone for their help and suggestions.

Link to post
Share on other sites
  • 0

I'm sorry to hear about the hardware failure, that's never fun to deal with. :(

 

And depending on how old that card is, I would see about RMAing it. Worst case, so you can resell it (as new, even) and recoup some of your cost (or just add it so you have more eSATA ports, etc).

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Answer this question...

×   Pasted as rich text.   Paste as plain text instead

  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.

Loading...
×
×
  • Create New...