What m/b and disk controllers are you using?  The 'spinup groups' feature was added back in the day when IDE drives were still common and SATA controllers were just 2-channel IDE controllers in disguise.  On an IDE controller, if you are streaming from one drive, say the master, and you spin up the slave drive, media pauses because the channel is hanging on the spinup.  This could also happen with older SATA ports.  Spinup groups lets you tag such pairs or sets of drives you want to spin up/down as a group.

 

Here I have this problem too. Maybe the shfs service, waiting other drive to spun up, freeze the ongoing stream. That's just a guess.

Here are a couple videos I shot to demonstrate the problem.  Please excuse the quality.  I took the video with my cell phone and it was hard to hold steady.  In both videos I am playing music that is on my unRAID server using J.River Media Center 17.

 

 

This first video shows multiple interruptions while running a library update in Ember Media Manager.  For those who don't know, Ember is software that scans the video files on your computer to create nfo files and various artwork for movies and TV shows.  The scan I performed did not create any files, it was just searching to see if any new video files were on my unRAID server and it did not find any.  It's best to skip to the 40 second mark.  I thought I clicked the button to start scanning right from the get go, but I must have missed the button and it took me few to figure out nothing was happening.

 

http://www.youtube.com/watch?v=9Lj-ISONmBo

 

This second video shows me copying a file to disk17 on unRAID.  Again you can here a cuople interruptions in the playback of the music.

 

http://www.youtube.com/watch?v=4VVdYb4PsF4

 

 

Here are a couple videos I shot to demonstrate the problem.  Please excuse the quality.  I took the video with my cell phone and it was hard to hold steady.  In both videos I am playing music that is on my unRAID server using J.River Media Center 17.

 

 

This first video shows multiple interruptions while running a library update in Ember Media Manager.  For those who don't know, Ember is software that scans the video files on your computer to create nfo files and various artwork for movies and TV shows.  The scan I performed did not create any files, it was just searching to see if any new video files were on my unRAID server and it did not find any.  It's best to skip to the 40 second mark.  I thought I clicked the button to start scanning right from the get go, but I must have missed the button and it took me few to figure out nothing was happening.

 

http://www.youtube.com/watch?v=9Lj-ISONmBo

 

This second video shows me copying a file to disk17 on unRAID.  Again you can here a cuople interruptions in the playback of the music.

 

http://www.youtube.com/watch?v=4VVdYb4PsF4

 

I have the same problem.  Though for me it can happen even with all disks spun up.

I see this also.  I use JRiver v18 for media playback.  I'm pretty sure it's an unRAID issue, but I can't rule out JRiver, since we both use it, although different versions.

Here are a couple videos I shot to demonstrate the problem.  Please excuse the quality.  I took the video with my cell phone and it was hard to hold steady.  In both videos I am playing music that is on my unRAID server using J.River Media Center 17.

 

 

This first video shows multiple interruptions while running a library update in Ember Media Manager.  For those who don't know, Ember is software that scans the video files on your computer to create nfo files and various artwork for movies and TV shows.  The scan I performed did not create any files, it was just searching to see if any new video files were on my unRAID server and it did not find any.  It's best to skip to the 40 second mark.  I thought I clicked the button to start scanning right from the get go, but I must have missed the button and it took me few to figure out nothing was happening.

 

http://www.youtube.com/watch?v=9Lj-ISONmBo

 

This second video shows me copying a file to disk17 on unRAID.  Again you can here a cuople interruptions in the playback of the music.

 

http://www.youtube.com/watch?v=4VVdYb4PsF4

 

I don't know if there's an issue with UnRAID or your hardware here, but my immediate reaction was "what is wrong with your player?  why is it not buffering?"  Yes, there may be something that can be improved in UNRAID, but I don't understand how any modern decent player does not automatically or by configuration buffer an A/V stream, so that it can maintain a constant playback speed.  Buffering has always been a fact of life, when you can't count on a constant source or have a non-local source, such as networked or from the Internet.  I don't know JRiver, but there must be some sort of configurable buffering.  When your source download speed is close to playback speed, then buffering can still have issues or you deal with the tradeoff of a long delay before start of playback.  But when it's a fast download source, such as from a file server, then the player should easily be able to build a full buffer in the background, with almost immediate start of playback.  Seems like 5 to 15 seconds buffered should be enough to guarantee smooth playback, no matter what happens to your source.

 

I can remember even portable (better quality) CD players had something they called 'anti-skip' (or something like that), which was a form of buffering.

Here are a couple videos I shot to demonstrate the problem.  Please excuse the quality.  I took the video with my cell phone and it was hard to hold steady.  In both videos I am playing music that is on my unRAID server using J.River Media Center 17.

LOL, I like it.  Nice choice in music too btw!

 

This first video shows multiple interruptions while running a library update in Ember Media Manager.  For those who don't know, Ember is software that scans the video files on your computer to create nfo files and various artwork for movies and TV shows.  The scan I performed did not create any files, it was just searching to see if any new video files were on my unRAID server and it did not find any.  It's best to skip to the 40 second mark.  I thought I clicked the button to start scanning right from the get go, but I must have missed the button and it took me few to figure out nothing was happening.

Is your music player accessing files via disk share or network share?

As Ember runs, do any disk spin up as a result, and if so, do you think your music skipping might correspond to that?

 

This second video shows me copying a file to disk17 on unRAID.  Again you can here a cuople interruptions in the playback of the music.

The music pause seems to overlap time where disk17 is probably spinning up, is that right?

After that initial music pause, as the copy proceeded, did music pause again?

If disk17 is already spun up, and you start the copy fresh, does music pause?

1.  Thanks.

 

2.  I'm not really sure the difference.  I initially pointed the software to the user share "Music" I set up in unRAID and it scanned in all the files into it's library.  The problem also happens If I navigate straight to the music file via Windows Explorer and play it with VLC.  When I run Ember some disks spin up, but not all.

 

3.  I think so.  It does seem to skip when I would expect the drive to spin up.  If you notice on the video, there is a second skip about a second after the first, but after TerraCopy shows some data has already been copied.  If I spin up all the drives and try either of my scenarios, I don't get any skipping.

I'd like to add what I've noticed in regards to interruption of streaming while writing to the array.

 

For me, interruptions of streaming appear to be network related, not unRAID related.

 

When I had a gigabit network, I could write from multiple sources, and stream to multiple clients with no problems.

 

I moved recently, and have temporarily set up MoCa to get my network everywhere easily.  Now I experience intermittent interruptions when writing to the array, and streaming to one single client.

 

Writes happen from a PC on the same gigabit switch as my unRAID server, so it's not that writes are using up all my MoCa bandwidth.

 

I guess with the MoCa networking, I just don't have enough spare bandwidth to fill my buffers and keep them full.  It's also extremely noticeable when skipping ahead during playback.  What before was an instantaneous jump now takes a bit to start playing.

 

This is all to be expected with as dramatic a reduction in networking I have had.

 

Maybe something to look into for people experiencing the problem?  It may not be the case for everyone, but it certainly appears to be in my case.

 

Also of note, I'm no 5.0-beta14, not one of the RCs.

 

2.  I'm not really sure the difference.  I initially pointed the software to the user share "Music" I set up in unRAID and it scanned in all the files into it's library.  The problem also happens If I navigate straight to the music file via Windows Explorer and play it with VLC.  When I run Ember some disks spin up, but not all.

 

 

Did you 'map' your 'user share' as a windows folder?  If you did, it would show up as a drive under 'Computer' in Windows explorer.  For example, I have mapped a share that shows up with a 'disk drive with a network symbol' icon in front and followed by:

   

      PIMAX Data (\\ELSIE1\Data\SharedData) (Z:)

 

The Z: is the drive letter that I assigned when I setup the mapping.  Any program can access this share as Z: 

 

(As a complete aside, I have had issues with VLC not handling networked sources gracefully.  Since I normally don't use it for that I have never really investigate it...)

I do not have any mapped locations.  Another note on media player buffering... that won't solve the problem either as far as I can tell.  The problem being that that interruption in playback varies greatly in duration.  I've seen anywhere from a fraction of a second to up to 18 seconds that I've timed.  Just now I timed one at 12 seconds long.  J. River's max buffer is one second, so that won't help much because most that I experience are over a second long.

I exclusively use mapped drives in windows, and have every thing connected via gigabit Ethernet, and still suffer this issue, FYI.

I do not have any mapped locations.  Another note on media player buffering... that won't solve the problem either as far as I can tell.  The problem being that that interruption in playback varies greatly in duration.  I've seen anywhere from a fraction of a second to up to 18 seconds that I've timed.  Just now I timed one at 12 seconds long.  J. River's max buffer is one second, so that won't help much because most that I experience are over a second long.

 

I'm not saying there is not a problem, since obviously 18 seconds, and even 12 seconds, is ridiculously long, but it is disappointing that JRiver has a max of only a second.  If it is able to play streamed music off the Internet, then other users must have been screaming for a longer buffer.  I would check with their support and see if that max is really the max, or whether they may have a registry var you can increase.  Since they have the code in there to buffer one second, they should have no problem buffering almost any number of seconds, it's just a number, unless they have hard-coded it, which would be poor design.

I do not have any mapped locations.  Another note on media player buffering... that won't solve the problem either as far as I can tell.  The problem being that that interruption in playback varies greatly in duration.  I've seen anywhere from a fraction of a second to up to 18 seconds that I've timed.  Just now I timed one at 12 seconds long.  J. River's max buffer is one second, so that won't help much because most that I experience are over a second long.

 

I'm not saying there is not a problem, since obviously 18 seconds, and even 12 seconds, is ridiculously long, but it is disappointing that JRiver has a max of only a second.  If it is able to play streamed music off the Internet, then other users must have been screaming for a longer buffer.  I would check with their support and see if that max is really the max, or whether they may have a registry var you can increase.  Since they have the code in there to buffer one second, they should have no problem buffering almost any number of seconds, it's just a number, unless they have hard-coded it, which would be poor design.

 

I know you're not saying there isn't a problem.  You're just trying to offer up some workarounds.  That's cool.

 

FYI.  I double-checked and apparently the setting I was looking at in J.River wasn't the normal buffer.  It was the DirectSound output buffer (whatever that is).  The correct setting is what J.River calls prebuffering and it can be set to a max of 20 seconds.  VLC, on the other hand seems to have a max of 9999ms.

 

I should also mention that I experience an interruption in video playback in XBMC whenever SickBeard's post processing copies a video to my unRAID server too.  Just another example.

I fixed my iTunes pausing issue by upping the playback buffer:

 

defaults write com.apple.iTunes file-buffer-size 32000

I tried setting my buffer to the max and I still get the problem.  Any word on a possible fix for this Tom?

Ran into this, finally I know I am not the only one. If I try to access any spun down drive, all communications for a lack of better words comes to a halt until that drive spins up.

 

So if say a plex client (buffer set to 30 seconds) via AFP is playing a movie on disk#12 and via  a windows box I access any other disk that is spun down. The movie pauses.

 

If I am copying something to say the cache drive and go to access any array drive that is spun down from the same client or any other client via SMB, the copy pauses and will not continue until the spun down drive spins up. No matter how you slice and dice it, there is a freeze across unRAID until a spun down drive spins up.

 

Not using spin up group, global spin down is set to 30min.

All my 18 drives are on LSI controllers.

 

Ran into this, finally I know I am not the only one. If I try to access any spun down drive, all communications for a lack of better words comes to a halt until that drive spins up.

 

So if say a plex client (buffer set to 30 seconds) via AFP is playing a movie on disk#12 and via  a windows box I access any other disk that is spun down. The movie pauses.

 

If I am copying something to say the cache drive and go to access any array drive that is spun down from the same client or any other client via SMB, the copy pauses and will not continue until the spun down drive spins up. No matter how you slice and dice it, there is a freeze across unRAID until a spun down drive spins up.

 

Not using spin up group, global spin down is set to 30min.

All my 18 drives are on LSI controllers.

 

Odd, Plex should continue if the buffer is 30 seconds. As soon as I set buffer in iTunes/Plex to read ahead 30+ seconds my problems went away.

 

I still experience an issue in Finder or Windows Explorer if you access a folder that is on a spun down disk, but that cannot be avoided. You get that problem with internal disks...

I ran into this thread because of your post of how to set the iTunes buffer (Thanks for that!)

 

I do have to say with Plex on my ma cmini the pause is not as severe as with finder/windows explorer (is your connection to unRAID for the first time via a OS X client take a long time? I am on 10.7.3). Almost certain its something else, sort off.

 

Let me explain, when launching say a TV Show, within about 5 seconds of play it pauses (even if nothing else is accessing unRAID but this one client) for a 2-3 second period and then continues on. I was think that the Plex Server (on another mac), is reading the folder/file (on the unRAID server) and that is what causes the pause for the Plex client. Hope this makes sense. As I know the Plex server needs to check the media and verify it against the client that's requesting it, should detect that it support direct play and then the mac mini plex client should direct stream from unRAID not the plex server (feature only available to Plex on OS X via AFP, hence why AFP is important to us for unRAID).

 

 

 

Oh and last night I finally had sometime to watch a movie with my wife (its been months) so first time via unRAID 5.0RC11a, and the movie blipped (about 1 sec) every 30mins (2 hour movie), which is what my spin-down group is set to. That used to never happen before. I want to say Beta12a never had any of these issue for my mac clients. So I am MAD as hell with unRAID 

 

Hmmm, no I'm not seeing that behaviour at all. I'm running 10.8.3, Realtek network on both sides so nothing flash.

 

10.7 is a bit shit, upgrade to 10.8 if you can. My connects are very quick, though I've just done a CNID rebuild (see my post in the latest unRAID release post).

Hmmm, no I'm not seeing that behaviour at all. I'm running 10.8.3, Realtek network on both sides so nothing flash.

 

10.7 is a bit shit, upgrade to 10.8 if you can. My connects are very quick, though I've just done a CNID rebuild (see my post in the latest unRAID release post).

Thanks speeding_ant, I will try the CNID rebuild. Question on that, it is something you saw that lead you to rebuild CNID?

I added -l LOG_MAXDEBUG and this allows me to see whats going on cnid_metad in the syslog. As far as I can see everything is fine. Been running like this to test, what happens when array shuts down, restarts, mac clients connect, plex scraps, etc.. all looks well.

 

But I will rebuild just to see the difference, do I have to stop the array, or AFP before performing the rebuild? (I just saw u posted to stop AFP in the other thread, thanks)

 

As for OS X, I will update to 10.8.3 I guess then, I am no mac guru, do you recommend upgrading or fresh install?

 

Many thanks for your assistance and suggestions.

Speeding_ant, on DBD -rf I see:

  -r Rebuild volume:

      1. Sync CNIDSs in database with volume

      2. Make sure .AppleDouble dir exist, create if missing

      3. Make sure AppleDouble file exists, create if missing

      4. Delete orphaned AppleDouble files

      5. Check for directories inside AppleDouble directories

      6. Check name encoding by roundtripping, log on error

      7. Check for orphaned CNIDs in database (requires -e)

      8. Open and close adouble files

      Options: -c Don't create .AppleDouble stuff, only cleanup orphaned.

              -f wipe database and rebuild from IDs stored in AppleDouble

                  files, only available for volumes without 'nocnidcache'

                  option. Implies -e.

 

'-f' I have 'nocnidcache' set in AppleVolumes.default, otherwise when those apple double files are generated, mover always crashes trying to move them into the array...

 

Thoughts? I don't mind having all the apple double files, its just unRAID's mover always crashes trying to move them  :'(

 

WARNING

 

In order to be able to run -rf reconstructing the CNIDs in the database from the AppleDouble files, make sure you've run a -r rebuild sometimes before, where the CNIDs then would have been synched between database and AppleDouble files.

 

Also be careful about the option nocnidcache. Avoid this option if at all possible, because if prevents you from being able to use -f.

 

CNID background

 

The CNID backends maintains name to ID mappings. If you change a filename outside afpd( (shell, samba), the CNID db will not know and not reflect that change. Netatalk tries to recover from such inconsistencies as gracefully as possible. The mechanisms to resolve such inconsistencies may fail sometimes, though, as this is not an easy task to accomplish. E.g. if several names in the path to the file or directory have changed, things may go wrong.

 

If you change a lot of filenames at once, chances are higher that the afpds fallback mechanisms fail, i.e. files will be assigned new IDs, even though the file hasn't changed.

 

I have seen this behavior "If you change a lot of filenames at once, chances are higher that the afpds fallback mechanisms fail, i.e. files will be assigned new IDs, even though the file hasn't changed." Sabnzb, Sickbeard are Windows hosts working off unRAID SMB.

I would very much like to see 64-bit with this.  Also, (and maybe I have not found it yet), I would like to see an easy way to keep my plugin's (SAB, CouchPotato, SickBeard among others) updated when a new release hits.  Even an auto update for the unRAID release would be nice too.

...  Also, (and maybe I have not found it yet), I would like to see an easy way to keep my plugin's (SAB, CouchPotato, SickBeard among others) updated when a new release hits.  Even an auto update for the unRAID release would be nice too.

To be honest, for me, an auto update feature could possibly be the worst way to ensure a reliable system.  There's a good reason that some people on this forum still run 4.3, 4.5, 4.7 etc.  They take the view that if it ain't broke, don't fix it.  The resultant stability is worth much more than having the latest version with the latest gizmos .... and the latest (unknown) bugs.

Here's my thoughts.

 

If everything is good with the current RC version, except the slow write speed for certain motherboards, then I say it's ready for final, however, before you release it as final you should also have a solution for the slow write speed issue.  If that solution is a 64bit version then it needs to be released at the same time.  If someone upgrades to 5.0 final but gets the slow write issues then they will have to go to the 64bit solution but that version will have to be made sure that it fixes the problem, not just create the 64bit version and it not solve the issue.

 

There will always be bugs, nothing is bug free.  The slow write issue is an inconvenience to some but Unraid is a write once read many product, it's an issue that should be looked at but not that important in the big picture of what the product is designed for IMHO.