Could the delay be from disks spinning down? Investigate at cache-dirs in the User Customization forum.

I am running 4.7 I have never updated to the Beta version. I have an Intel PCI-e NIC that installed in the Unraid box without a hitch, and is immediately recognized. Everything works just fine, 1080p BDs being streamed to WMC7 x32 HTPC, cat6 & Gigabit switches, the jumbo frames I believe have nothing to do with Windows explorer and its funny behavior as this happened before I used jumbo frames, I was looking into jumbo frames to get rid of the problem with Windows explorer. I am copying from a QNAP NAS using TeraCopy 43Gigs at a steady 30MB/s straight to the array, no slow down here. I am copying from my local machine straight to the array between 42MB/s and 50MB/s the network is fine. I have a 64bit Win7 machine as workstation & a 32bit Win7 HTPC. The connectivity is fine and the speed is more then acceptable. The problem is Win Explorer GUI reading the shares and its folders. In rare occasions it stops in the middle of copying and gives the error "Destination no longer exists", horse pucky. I open another explorer window and have to egg the machine on to reach the destination. At times a new explorer instance displays the directories but the old explorer window that had the error message does not. I am thoroughly convinced that this is a Windows issue that was introduced in Vista and never fixed in Win7. All these people complaining about slow moving and copying using Win7 explorer can't be wrong. If I had an XP machine I would research it but it is too time consuming to get to the bottom of what is a Windows issue.  

 

As a new effort to get rid of the problem I ordered a new Gigabit switch "TRENDnet TEG-S80G 8-Port Gigabit GREENnet Switch" Connected only the Unraid server, workstation & HTPC to this switch plus a connection to the Gigabit switch that is connected to the 100MB/s Comcast router providing Internet access. There are 3 devices on this segment, HP Workstation Win7 x64, Desktop switch to which Win7 x32 machine is connected and Unraid server. There should be nothing inhibiting transfer & visibility between the workstation & the Unraid server. After a few days running this setup I'll see if this fixes it.

 

So far copied a 43Gig BD folder with this setup from NAS @ 30MB/s and no funny business, no early tell tales from TeraCopy with the destination not being available. Will see if this did the trick, only time will tell.

 

Just saw the post about spin down. Before I copy I always spin up the discs and I do use the cache-dir script to keep the directories in memory.

Well, I'm encouraged, and discouraged all at the same time.  your symptoms perfectly describe mine.  the intermittent unavailability, new windows show all fine, etc, all the same.

 

I haven't turned off jumbo frames yet, but will later tonight, just to eliminate that also.

 

I also have mine wired the same as you it sounds, but with one more computer connected via wireless, but have tested with it turned off also.

 

I thought I saw some thread about caching the directories several levels deep, but am not sure where I saw it (I've read and re-read so much already) or if it would even help.  It sounds like you've done this also, but it didn't help you

 

I've never ran anything but this beta version, and I do understand potential "weirdness" about beta, but it seems to be unrelated to that.  Is your Intel PCI-e card the same as mine, posted a few posts back?

 

I was doing some more testing last night, from the HTPC wired to the switch, and using JRiver Media Center v17 I renamed some 30GB mkv files, which caused them to be moved to a new folder on the same physical drive, and it was instant.  Literally, by the time I opened the correct folder in IE to verify when it finished, it was done.  I'm sure it just updated the record to reflect the new folder, and didn't physically move the file, but I was impressed that a windows program would act this way on the Linux drive.

 

Anyway, I suppose I'll also start casting a wider (inter)net to get some more ideas to eliminate this weirdness.  I would think more people would be complaining if they experienced this, so I'm still hopeful we can eliminate it somehow 

I had tossed this around should I or shouldn't I until I read your description of the problem and I said let me chime in perhaps this can get fixed. I am not very savvy with Linux but realize that these are two different worlds and having at one time made a living working with VAXs and trying to connect to them using Windows for Workgroups I came to realize that Unix was invented by space aliens and Windows tried to be an operating system in that world. I have the highest respect for the Unix world and its OS but the apps & eye candy is in the windows world and they don't have too happy of a marriage probably because of the cockiness of the MS world. That is what we average Joes are stuck with and can afford.

 

I will monitor my new setup carefully and see if this makes any difference. the devices in my network are as follows:

Connected to a D-Link DGS-2208 8-Port Desktop Green Ethernet Switch is a Win7 32bit HTPC. The switch is connected via cat6 to a TrendNet 8-Port Gigabit Switch. To this TrendNet switch are connected all GigaBit devices, UnRaid server with Intel EXPI9301CTBLK Network Adapter, HP-Pavillion e9280t Win7 64bit, QNAP TS-239 Turbo NAS and a OfficeConnect Managed Gigabit Switch - 3Com 3CDSG8 who provides the Internet access from the 100Mbps Comcast modem. This portion of my network hooked up to the TrendNet 8-Port Gigabit Switch should have nothing but performance goodness. 

 

Until this point in time it looks good so far with all the devices needing to interface with the UnRaid server being on one switch. Hope that this will hold for the rest of the week and maybe this is the answer to our problem, somehow I just can't shake the feeling that the problem is being masked by this approach, or it could be that not all devices play fair.

...the devices in my network are as follows:

Win7 32bit HTPC  ==> D-Link DGS-2208 8-Port Desktop Green Ethernet Switch ==> TRENDnet 8-Port Gigabit Switch ==>  UnRaid server (via Intel EXPI9301CTBLK NIC) & HP-Pavillion e9280t Win7 64bit & QNAP TS-239 Turbo NAS & OfficeConnect Managed Gigabit Switch - 3Com 3CDSG8 ==> Internet access from the 100Mbps Comcast modem.

 

A couple of observations.  it appears you have 3 computers hooked up to 3 gigabit switches and your modem.  I'm sure you have your reasons for having so many switches in your system for so few computers, but I might start eliminating some of them and see if that doesn't help.

 

I said earlier that it sounded like we had a similar setup, but that's not quite accurate after seeing your setup.

 

We both are using the same Intel NIC (which I just purchased based on good review/recommendations), and both go into a D-Link switch, but mine (DGS-1008G) is different than yours (DGS-2208)

 

However, my network is wired more directly.  I have a U-Verse 10/100 modem/router ==> D-Link switch ==> unRAID (using the same NIC) & HTPC & Yahama AVR.  A desktop also connects via wifi-g, but all machines are wired to the switch with brand new cat6 cables and to the router with a new cat6 cable also.

 

So, either we both have the same weird thing configured/wrong in windows (likely), or the NIC and Gig-e switch combo is problematic for both of us, perhaps related to jumbo frames, which I will turn off and test again without this evening.

 

I will look again when I get home to see if I can find the link/thread/FAQ/Wiki about caching directories, and see if I can get that setup also (separate from the jumbo frames test/change).

 

I also was looking at services in Windows the other day and saw some service that sounded like it might be related to this issue, but can’t remember what it was called, and got sidetracked before I investigated it, so I’ll look into that also.

 

I’m looking thru the services on my XP machine at work, and here are a few that sound promising.

 

Computer Browser - Maintains an updated list of computers on the network and supplies this list to computers designated as browsers. If this service is stopped, this list will not be updated or maintained.

Indexing Service - Indexes contents and properties of files on local and remote computers; provides rapid access to files through flexible querying language.

Network Access Protection Agent - Allows windows clients to participate in Network Access Protection

Network DDE - Provides network transport and security for Dynamic Data Exchange (DDE) for programs running on the same computer or on different computers. If this service is stopped, DDE transport and security will be unavailable.

Network DDE DSDM - Manages Dynamic Data Exchange (DDE) network shares. If this service is stopped, DDE network shares will be unavailable.

Network Provisioning Service - Manages XML configuration files on a domain basis for automatic network provisioning.

 

These were all turned off on the XP work machine, I’ll check these at home also.

 

Anyone know anything about these services, and if any should be running to help alleviate this issue?

 

moved the Yamaha AVR to connect directly to the modem/router, since they are both only 100Mb/s. didn't help.

 

I had not added the line to the go script to make the jumbo frames sticky, so a reboot cleared that.  didn't help.

 

read around, stumbled on this...

 

run an "elevated" command prompt (as Admin) and copy the following

 

netsh int tcp set heuristics disabled

 

right-click in the command window, paste, enter

 

go check your folder access.  I haven't had any pauses on 2 windows7x64 machines in 45 minutes since doing this.

 

I did more settings on my desktop, then checked, but on the HTPC, this was the first one I tried, and it's been great so far.

 

Curious to see if it sticks overnight, but it has already stuck thru a restart on the HTPC.

 

Hopeful 

My reasoning to set this network up was because of location. I have 3 rooms in the house the Home Theater room is on one end of the house upstairs the UnRaid server and QNAP NAS and Internet Access entry point is in a different room also upstairs but on the opposite end of the house. I run from my Trendnet GigaBit switch about 80 to 100 feet of cat 6 cable to connect to the D-Link DGS-2208 switch. The D-Link switch has two connections the HTPC & my Denon 4311ci receiver. This is normal and not problematic because just 1 network cable to serve the HT room. The Trendnet switch has 3 devices UnRaid Server, QNAP NAS & HP workstation which is ground floor office. So far for the past two days it has been working excellently. Just transferred from the QNAP NAS to my HP Workstation with excellent results and also transferred from HP Workstation to UnRaid server with 40MB/s to 49MB/s.

 

Under my old setup when using TeraCopy I often have for the files waiting to be copied, a red icon all the way in the left column, indicating that the location can't be found or is not available while the ones copied have a green check mark and the one that is being copied has a green arrow head. Then it moves to the next file and normally it starts to copy but every now and then it gets into the mode that we are trying to fix. When using the normal Windows copy method and it just sits there with the progress bar just not moving, then when clicking the button that says "More Details", it displays the message "calculating" for "Items Remaining" & "Time Remaining" as it is reading the file and is trying to sort out the destination. That is why I believe this whole thing is a Windows issue. If you connect a XP machine to your network and experience the same issues on the XP box then I stand corrected and it is a network or network configuration issue.

 

Under my new setup so far with TeraCopy I do not see the red icons for the files waiting to be copied rather they have no icon, I attach a screenshot of it. For the past two days I am pleased with the UnRaid server performance in this modified setup. It could be that the 3rd switch the 3Com 3CDSG8 since it is a Managed Gigabit Switch might be the cause of the problem. The TrendNet switches with Hardware version 2.0 for $30 was an excellent deal and I hope that the isolation of all devices and parts that have nothing to do with streaming content from UnRaid will solve our  problem. Again only time will tell and I hope with shared knowledge and research we get this thing licked, especially if it is a Windows issue or some Network tweaking.

 

Your findings are very interesting after Googling for that command I landed on this page:

Speedguide.net - Windows 7, Vista, 2008 Tweaks

 

Let me know how things are turning out over the next few days. I will see how this "Disabling of Windows Scaling heuristics" modification will work on my end. Perhaps some of the network spezies here could shed some clarification on this.

Your findings are very interesting after Googling for that command I landed on this page:

Speedguide.net - Windows 7, Vista, 2008 Tweaks

 

Let me know how things are turning out over the next few days. I will see how this "Disabling of Windows Scaling heuristics" modification will work on my end. Perhaps some of the network spezies here could shed some clarification on this.

 

Yep, that's where I found it.  I applied/changed much of what's on that page, and no issues for the last couple of hours.  I have gone back and forth between the 2 computers and have not had a single instance of crazy lag (or any really).

 

It takes 5 seconds to do, and the results are immediate (I sound like an infomercial)

 

I think this eliminates one of the big niggling things from my life

 

I'll worry about overall transfer speeds again a bit later.  At least this is usable, and as fast as it's ever been actually, in displaying folders/contents.

 

if I don't get any regression in the next day or so, I'll update to post solved

run an "elevated" command prompt (as Admin) and copy the following

 

netsh int tcp set heuristics disabled

 

right-click in the command window, paste, enter

 

go check your folder access.  I haven't had any pauses on 2 windows7x64 machines in 45 minutes since doing this.

 

It's been about 6 hours of actual use on this machine and another hour or so, plus a few hours of TV watching on the HTPC, and each machine has stalled out/hung up (clicking a directory took 45-150 seconds to become responsive) twice in all that time.  I'm sure all related to the array spinning up.  Otherwise, any browsing of any directories or folders has been very fast.

 

The step(s) above is the ONLY thing I did on the HTPC, and it's been great.  I admit I am astounded that switching a 1 to a 0 made this kind of difference.  I'd call it snake oil, but whatever, it's working great here.  hard to believe this *isn't* the default 

 

Network transfer speeds could improve, but with this functioning, I can move on to worrying about that (and getting my desktop wired for cat6).

 

Also, on the desktop, I did run and apply the suggestions from this tool, but it was mostly a GUI for the dos commands I had already done.

 

SG TCP Optimizer from http://www.speedguide.net/downloads.php

During the past 2+ days I observed that the separating of the devices that interface with the UnRaid server virtually eliminated the problem of unresponsiveness. However the strange things that pop up in dialog boxes as well as in the TeraCopy display are still there. Disabling of Windows Scaling heuristics did not make the night and day difference that Justin Chase experienced. However I felt that keeping it enabled gives it just a little edge in speed transfer like 36MB/s instead of 34MB/s but nothing to write home about especially if it solves the hanging and unresponsiveness of the folders and drives.

 

Here is the reason why I believe that it is not 100% sorted yet.

1. When I tried to delete a 21.3 gig folder I expect it to be almost instantaneously not sitting there telling me that it discovered 41 items making up 21.3 gig and now it is waiting to calculate the time remaining & Speed. Just zap the damn thing without the GUI fluff. When I want the GUI fluff I'll ask for it. Must be a Windows thing, anyone knows if that can be turned off.

 

2. When copying a 712Meg file the 2k NFO file is shown as not being able to be moved because the system can't find the destination drive. WTF it is about to move a file there "0.4% done". This was my first indication that something is not kosher and it is still there, that is why I believe it is not sorted and in time perhaps it will rear its ugly head again.

 

3. When finished copying it did find the destination, why the red "X" icon why not a blank space as shown three posts above. Beyond my knowledge of Linux & Windows network interface & low level OS operation. Also note moving 712 Meg in 1 minute & 16 seconds not the fastest because it set there waiting and thinking what next.

 

Hopefully I will not have to return to this thread. Now on to the next job getting the Supermicro 8 port controller to work with version 5B12a.

I think I finally liked this problem. I implemented the suggestion in this thread and made changes to the Windows Credential Manager. I added the Tower IP address together with the User root and its password. This seemed to make all transfers and navigation instantaneously and snappy. We will see what it looks like in a week. Hope your rig is still working as intended.

I think I finally liked this problem. I implemented the suggestion in this thread and made changes to the Windows Credential Manager. I added the Tower IP address together with the User root and its password. This seemed to make all transfers and navigation instantaneously and snappy. We will see what it looks like in a week. Hope your rig is still working as intended.

 

Unsurprisingly, it's come back, and I've spent the last week or so messing with all kinds of network settings, and some days it's better, and today, i cannot connect to the unRAID server at all, from only one machine.  nice.

 

I'll try the suggestions in that thread, and some others I've found also.

 

Love Microsoft!!!

I was never quite convinced that we had licked this problem. Deep down I had the gut feeling this has to do with network idiosyncrasies because of Win7 & Linux, plus perhaps permissions or SAMBA, NFS and whatever else is needed to get these two OSs to play fair with each others. So far I noticed that two suggestions have helped a little bit to make it more predictable. The change of the switch to an unmanaged switch and isolating it from everything else was an uptick. The change to Disabling Windows Scaling heuristics was also an uptick but nothing solved the problem totally, just an ever so slight improvement to deal with the problem. However the latest change of including the credentials in the Windows Credential manager looks perhaps the most promising change so far. I really wish this annoying behavior would disappear.

Did you guys ever find an answer to this?  I am having a very similiar issue.  I have replaced NIC cards in the server and the PC's, replaced the switch and followed the solutions here and none of it seems to work for long.

 

It is to the point where unRAID is almost useless.

 

 

Mike

No the problem has never fully disappeared, however in my case it is manageable and the annoy factor is bearable. I am using teracopy which seems to work better then explorer type copy or move. When you use the copy or move with Windows Explorer the dialog box just sits there and the green progress bar just sits there and indicates it is scanning. When you click the more button the 3 parameters that show size, remaining time and what has been copied just sit there displaying "Calculating". WTF does it need to calculate I don't want to know that just be off copying or moving & I don't care about how much time is left and how much has been moved/copied. It will be done when it is done, stupid idiosyncrasy.

 

Anyhow since the moving/copying is not on the same level as I do on my desktop I can grin and bear it.

 

However I have seen a big improvement using the HP micro server running four 3TB drives AND using version 5b14 with the shares set to public access. The micro server is much better then my other Unraid server running 4.7 having movie, documentary & music video shares. Since the micro server with its TV shares is much more active in copying / moving I can handle my situation at the moment. Hope you get yours sorted, have you tried version 5 beta14, it has been rock solid and a pleasure for me.

When you use the copy or move with Windows Explorer the dialog box just sits there and the green progress bar just sits there and indicates it is scanning. When you click the more button the 3 parameters that show size, remaining time and what has been copied just sit there displaying "Calculating". WTF does it need to calculate I don't want to know that just be off copying or moving & I don't care about how much time is left and how much has been moved/copied. It will be done when it is done, stupid idiosyncrasy.

 

Often known as the "green ribbon of death" - widely reported in Vista and Win 7 but few reliable fixes.

Often known as the "green ribbon of death" - widely reported in Vista and Win 7 but few reliable fixes.

Thanks for that bit of information had no idea of that terminology. I have read plenty about complains of copying speed but "Green Ribbon of Death" so appropriately describes what is happening. I have lost moving data because it hung on this dialogue box than came with a message that the destination is no longer available and the data was gone. So I quit moving and started copying and then a separate delete. However I have yet to experience the problem with version 5 beta 14. Deep down I always knew that this was a windows problem not playing nice with Linux.

 

But thanks again as now I have a Google term that I can search for.

 

EDIT:

Found this in a document and now I know what is happening. It is not UnRaid it is bleeding Windows, so don't blame UnRaid but windows for creating the "Green Ribbon of Death" the cousin of the "Blue Screen of Death".

Take the Blue Screen of Death, a Windows mainstay for more than a decade. Yes, it’s still alive and well in Windows 7, but now it has a cousin: the Green Ribbon of Death. As explained in Chapter 2, the Green Ribbon of Death—capable of bringing Windows Explorer to its knees—comes from a combination of poor design and bugs in its code. And thus the reason for distinguishing where an annoyance becomes clear: you need to know what you’re dealing with in order to fix it.

 

Copying files

Windows 7 is hopelessly slow at copying files in certain situations, two in fact.

First, the UAC feature forces Windows Explorer to evaluate the security impact of each file you copy, and this has far-reaching consequences, particularly when you’re copying files over a network.

 

And second, Windows 7 notoriously has trouble copying files to and from USB devices. So, if you copy a folder full of images from your USB card reader directly to your external USB hard disk, or move document files from a USB flash drive to a shared network folder, Windows Explorer may crash. There’s no easy fix to this one, but you can work around it by copying files to your desktop first.

 

You need to use Task Manager to close a crashed Windows Explorer window. But if you want to be able to close a crashed window and leave any other Explorer windows (and the desktop) intact, you’ll need to make a change in Control Panel. Open Folder Options and choose the View tab. In the Advanced settings list, turn on the Launch folder windows in a separate process option, and then click OK. From now on, when you see the Green Ribbon of Death, it’ll only mean death for one of your Windows Explorer windows, not all of them.

I am wondering if Windows Explorer is scanning the entire file system on the server for some reason before it mounts it for use.  I know it scans USB keys and asks if you want to 'play' media files every time you plug a key in. 

 

Other item to consider: Is windows attempting to index all of the files on the server?  (See Control Panel, Indexing Options)

 

I see the same problem on my FreeNAS server where I have a lot more files than I do on my unRAID test bed. 

 

Just some food for thought...

Don't know about the indexing, but another suggestion is to change the folder options to always show icons and never thumbnails. Perhaps the problem comes from generating thumbnails but I am not sure, it is a windows thingy.

Don't know about the indexing, but another suggestion is to change the folder options to always show icons and never thumbnails. Perhaps the problem comes from generating thumbnails but I am not sure, it is a windows thingy.

 

I did find a post which did say if a file is corrupted so that it can't be opened to generate the thumbnail, it will crash Windows Explorer!  It might be well to turn off this 'feature' in Windows----  Control Panel, System, Advance System Settings, Performance--Settings, Visual Effects, Uncheck 'Show Thumbnails instead of Icons'.