Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[flac.rules]

When an operation takes a long time, please implement some kind of progress bar or "working on it", or something, instead of the whole program seemingly hanging up. (I have no clue if it takes less than an hour)

[glynor]

I split your request because your request is quite generalized, and I strongly suspect that there is something else going on.  I've seen other posts from you about hangs as well, so I thought it was worth discussing.

To be clear: MC makes extensive use of multithreading and background worker processes.  There are very few instances in MC where the application hangs because it is doing something that is expected to take a long time.  There are a couple individual exceptions I can think of (where things should be moved to a background thread with a progress bar in the Action Window).  However, you didn't explain specifically what instances cause these hangs, which leads me to believe you have something else going on.

So... Can you explain the types of situations where you are experiencing these hangs?  I'm not trying to deflect, but... What Jim was looking for in that other thread is, for example:

MC hangs when you use drag-drop from a file listing to copy files to Windows Explorer (or the Finder), while the copy process completes.  It would be better if this operation was done in a background thread so that MC itself can remain responsive.

(In fact, I'll add that to the thread, because it is a good one.)

But generic "it hangs a lot an you should make it not do that" isn't really going to help, and the issue might be something with your system or settings specifically anyway.  Understand what I mean?

So... Can you give us specific examples of when MC hangs?  If they are things others can reproduce from individual, repeatable actions, then we can post suggestions to fix those to the Too Easy thread.  If not, I can probably help you get some of them sorted out.  If you haven't read this:
http://wiki.jriver.com/index.php/Troubleshooting_Guide

I wrote that specifically to collect a bunch of the knowledge I've gained over the years in combating small hangs and non-responsiveness in MC.  I can guarantee, if things are working correctly, MC does not hang often.  If it is for you a lot, seemingly randomly, then something else is going on with your system.

[6233638]

To be clear: MC makes extensive use of multithreading and background worker processes.  There are very few instances in MC where the application hangs because it is doing something that is expected to take a long time.  There are a couple individual exceptions I can think of (where things should be moved to a background thread with a progress bar in the Action Window)

It happens on a regular basis for me every time I either insert a new optical disc, or try to play a disc which has spun down.
The former can sometimes be frozen for several minutes at a time while it "scans" the disc.
It's considerably worse if I'm already ripping a disc in MakeMKV, and insert or try to access a disc in another drive at the same time.
 
If my internet connection is busy/slow, trying to download a video off YouTube freezes the program for maybe a minute until the download prompt appears.
 
Views using IsEmpty() often take a very long time to evaluate, with no way to back out or cancel until it runs.

[flac.rules]

To clarify, my request was not too fix all instances which causes a hangup, my request was to make the program behave differently when this instances occur. I don't know if it is easy, as I am not a programmer, but when the program uses 100% CPU and the UI does nothing, the program should detect it, and somehow show that it is in fact working on something, whatever the root cause of the problem is.

As for my specific problems, what causes me most problems seems to be that MC in some ways poorly handles the Client-server-setup when it comes to performance. The firewalls are open, AV does not check MC and the network is fast, yet some operations are slow. I think it is a problem with JRMCs handling of stuff, because if i just map up the drive, and add files locally, some of these problems disappear, even though the network operates at the same speed (1000 Mbps).

I am actually compiling a list of these things after the last thread, to give some specific examples.

[mwillems]

It happens on a regular basis for me every time I either insert a new optical disc, or try to play a disc which has spun down.
The former can sometimes be frozen for several minutes at a time while it "scans" the disc.
It's considerably worse if I'm already ripping a disc in MakeMKV, and insert or try to access a disc in another drive at the same time.
 
If my internet connection is busy/slow, trying to download a video off YouTube freezes the program for maybe a minute until the download prompt appears.
 
Views using IsEmpty() often take a very long time to evaluate, with no way to back out or cancel until it runs.

+1; pretty much anytime MC is doing disk i/o/network access and it takes slightly longer than expected I get a hang that makes MC nonresponsive.  That includes:

1) New optical disk inserted / optical disk spin up.
2) Copying files, especially when copying to network drives. This is true for me even when using MC's rename/copy tool if remote drives are involved.
3) Situations where an MC client is playing and the Wi-Fi is interrupted (MC will show buffering, but any attempt to interact with MC causes MC to hang for minutes or sometimes indefinitely).
4) And certain features that require internet access: trying to get artist images from last.fm for example, always pops up a modal dialog that hangs MC if I'm doing more than two or three artists at a time.

[glynor]

It happens on a regular basis for me every time I either insert a new optical disc, or try to play a disc which has spun down.
The former can sometimes be frozen for several minutes at a time while it "scans" the disc.

This is a limitation of Windows, unfortunately.  The whole filesystem of Windows locks in these instances.  It applies to disconnected Network volumes as well, and all sorts of other "disk access" issues to sleeping disks.

Windows is less brain-dead about it than OSX (where the whole filesystem is single-threaded), but it is still pretty brain-dead.

If my internet connection is busy/slow, trying to download a video off YouTube freezes the program for maybe a minute until the download prompt appears.

Yep, this could be better, I think.

Views using IsEmpty() often take a very long time to evaluate, with no way to back out or cancel until it runs.

The View model isn't designed to be cancelled.  You may want to re-evaluate your expression, or further filter the list before doing the "expensive" calculation.  I have a couple views that are slow to load (my All Media views and the ones that do IsMissing() for example) but I try very hard to not use complex, expensive expressions in everyday views.  You can tune this, in other words.

Filters are evaluated in order.

[glynor]

To clarify, my request was not too fix all instances which causes a hangup, my request was to make the program behave differently when this instances occur. I don't know if it is easy, as I am not a programmer, but when the program uses 100% CPU and the UI does nothing

The instances where this happens (and it actually pegs the CPU) should be incredibly rare.

That is a strong indicator that something odd is going on with your system, and in all likelihood, it is something covered in my Troubleshooting Guide.

[glynor]

3) Situations where an MC client is playing and the Wi-Fi is interrupted (MC will show buffering, but any attempt to interact with MC causes MC to hang for minutes or sometimes indefinitely).

I agree on this.  It isn't specific to WiFi or anything else though.

MC could have much better feedback and responsiveness when buffering content from a Library Server.

[6233638]

This is a limitation of Windows, unfortunately.  The whole filesystem of Windows locks in these instances.  It applies to disconnected Network volumes as well, and all sorts of other "disk access" issues to sleeping disks.

OK, but is it not possible for Media Center to run I/O operations in a separate process/thread so that it doesn't lock the rest of the program with it?
 
(I'm sure it's just that easy! )
 
Even if you still couldn't interact with the program, preventing Windows from graying the app window and suggesting that it's frozen would be a big improvement.
 
I've often had MC in a situation where it has frozen like this, but I still had a video playing in a split pane which was fine until Windows decided to gray the app window because it appeared to be frozen.
 

The View model isn't designed to be cancelled.  You may want to re-evaluate your expression, or further filter the list before doing the "expensive" calculation.  I have a couple views that are slow to load (my All Media views and the ones that do IsMissing() for example) but I try very hard to not use complex, expensive expressions in everyday views.  You can tune this, in other words.

It's just frustrating if I open the wrong view by mistake. It's only certain maintenance views which normally use this expression.
 
It's probably a lot more frustrating as well since I use a split view, and could be dealing with things in the other pane while it checked to see if all the files on all my drives are there, if it didn't freeze the program when doing it.

[AndrewFG]

In my own programming now, when I write say a thread to do something that could hang, I also try to write a second "heartbeat" thread class that monitors the first thread, and if it detects no "pulse" in the first thread, it tries to unload the first thread. For example if the first thread hangs on an Http socket call for more than a certain time, then the heartbeat thread will kill the socket and force the first thread to terminate.

[glynor]

OK, but is it not possible for Media Center to run I/O operations in a separate process/thread so that it doesn't lock the rest of the program with it?
 
(I'm sure it's just that easy! )
 
Even if you still couldn't interact with the program, preventing Windows from graying the app window and suggesting that it's frozen would be a big improvement.
 
I've often had MC in a situation where it has frozen like this, but I still had a video playing in a split pane which was fine until Windows decided to gray the app window because it appeared to be frozen.

Unfortunately, not.  The issue is that it is Windows Explorer that is frozen, not MC, but Windows Explorer is the shell, so when it is frozen, everything that depends on disk is frozen.  Windows does not provide the interrupts needed to "escape" from this state when dealing with disk latency.  That's why basically every application you've ever used suffers from these same kinds of issues.

There might be some small things they could do around the edges to improve it, but the fundamental issue is with Windows.

As I said... It could be worse.  The OSX filesystem threading is a disaster.

[glynor]

I also try to write a second "heartbeat" thread class that monitors the first thread, and if it detects no "pulse" in the first thread, it tries to unload the first thread.

MC uses a message pump multithreaded design, which is pretty much SOP for multi-threaded applications.  Matt has discussed this before on the forum.

[glynor]

I should note, if you are having issues with this kind of problem, and it seems particularly tied to disk performance, and you haven't read this entire wiki article, you are doing yourself a great disservice:

http://wiki.jriver.com/index.php/Troubleshooting_Network_and_Slow_Storage

In particular, this can make a huge difference:

Options > Tree & View > Advanced > Display missing file image in lists: Disable

You may also want to experiment with disabling sleeping on your main Media Drives.  This is also a common cause of hangs in all applications (spinning up disks hangs the calling application).  And, unless you commonly go full days without spinning up those disks even once, the power advantage to parking them is going to be very limited, and the repeated parking/spin-up may shorten the lifespan of your drives.

Drive sleeping is a decent feature on Laptops where every watt-hour counts (well, actually, you should just get rid of spinning disks on Laptops), but it usually doesn't make much sense on a desktop/server machine that you use every day.

[flac.rules]

The instances where this happens (and it actually pegs the CPU) should be incredibly rare.

That is a strong indicator that something odd is going on with your system, and in all likelihood, it is something covered in my Troubleshooting Guide.

Ok, skip the 100% CPU part then, when the UI hangs completly, the program should behave differently, its not user friendly, and other programs very seldom have the same problem.

[adlelare]

623 and others,
i've have a similar very frustrating experience with JRiver hanging constantly for some period of time (with similar sort of experience).  on the weekend i had OTA TV playing on one laptop via JRiver for hours with no problem but this instance does not serve any other computer.  the TV was via HDHR (networked) so it got me to thinking, i then unhooked my primary HTPC with JRiver (media server) from my network and used my HP950Q tuner (direct to the PC) and i have been watching TV perfectly now for hours... no hanging no spinning wheel...hmmmm.. i will later be testing a video, some songs, etc., then rehooking the network but disabling media server capability o see what happens ( i suspect i will get perfect playback)...

[flac.rules]

Here is an example of a concrete situation of what gives me trouble:

Server A: Has a MC 20 library server running, no AV, firewall open, 500 MB/s disk-array and gigabit net.
Server B: Has a MC 20 client, connected to server A, no A, firewall open, gigabit net, 4 sound-cards which is sent to my speakers in different rooms in the house.

Both have Windows server 2012 R2.

If i run audio on server B, it can behave normally, and often will in the start of playback, most of the time it runs at 25% CPU (it should be less than 1%) on a quad core xenon, and it will quite frequently hang up the UI totally, when it does, audio keeps playing normally, until the song ends, and if the UI still isn't responding, there will be no sound for up to a minute, when the program reacts again, and jumps to the next song.

Capture 1 shows the UI hanging up, and the CPU usage, capture 2 is the situation a couple of minutes later when the program has recovered. Capture 3 is the network traffic, which is low, with a slight peak when the song is changing.

[JimH]

I notice you're running Windows Server.  Are you also modifying the OS with someone's special tweaks? 

[flac.rules]

I notice you're running Windows Server.  Are you also modifying the OS with someone's special tweaks? 

No, server B is as close to a vanilla install as you can get, it only has a couple of Oses that are viritualized (not running at the moment of the test), and MC installed, and basically nothing else. Server A doesn't have any special tweaks either, just regular server 2012 R2 essentials.

[gvanbrunt]

I'm curious what process is using up 25%. Because that is basically one thread maxing out a core. You didn't say specifically if that was MC or not.

[flac.rules]

It is MC that uses the CPU. The load is pretty evenly distributed between cores, no single core is maxed out.

[kstuart]

You may also want to experiment with disabling sleeping on your main Media Drives.  

In some cases, I use USB External Hard Drives, and the usual methods don't work for that.  I'd love to find some method that works for all of those.

[Hendrik]

It is MC that uses the CPU. The load is pretty evenly distributed between cores, no single core is maxed out.

Don't trust the task manager, even a simple task that is not multi-threaded and using 100% of one core will not show up as such in the task manager. Its deceitful!

[flac.rules]

Don't trust the task manager, even a simple task that is not multi-threaded and using 100% of one core will not show up as such in the task manager. Its deceitful!

Ok, maybe the task manager lies about how the different cores are being loaded, I don't know, the problem is still there though.

[gvanbrunt]

Don't trust the task manager, even a simple task that is not multi-threaded and using 100% of one core will not show up as such in the task manager. Its deceitful!

I know Hendrik knows this but for other that may not:

In addition just because "MC is using 25%" does not mean it's the cause of the 25%. Every program relies on the OS for many functions and if any of the them max out, so does the use for the program that called them. And a majority of the time time the "bug" is located it is with an underlying piece of hardware or it's driver. The trick is narrowing down the setup so we can figure out what exactly is causing the issue. Not that 25% is necessarily unusual by any means depending on what settings you have enabled etc. But it seems to me in your case something is maxing out a core as it "continues to play" which is on another thread while the UI thread is hung. So after it finished you can't do anything.

MC is multithreaded and doesn't often do blocking work on the UI thread (there are some exceptions to this). So when something hangs, it is usually due to a call to the OS that the UI is waiting on. Something that normally isn't blocking, but is in the case of this particular system. In most cases I've seen it happens to be drivers or hardware issues. So anything you can provide that might point to the culprit would help.

[JustinChase]

Today, I was doing some cleanup work, which involved rename, move, copy files, and update tags from library.

If you run update tags from library on a couple thousand files, mc doesn't 'freeze' but it really runs slow, and several 'functions' don't work at all.  ctl-a (select all), for example doesn't work while it's updating in the background.  You can click edit, select all, but ctl-a doesn't work.

Anyway, while update is running, or files are being renamed, MC just runs really slow.  Plus, the dialog showing the update tags gets 'hidden' behind MC, so you can't even see it's progress, and if you're not aware of the issue, it's really frustrating because you just don't know what's going on.  There should at least be a progress dialog, preferably with the option to pause and/or cancel the task(s).

[flac.rules]

I know Hendrik knows this but for other that may not:

In addition just because "MC is using 25%" does not mean it's the cause of the 25%. Every program relies on the OS for many functions and if any of the them max out, so does the use for the program that called them. And a majority of the time time the "bug" is located it is with an underlying piece of hardware or it's driver. The trick is narrowing down the setup so we can figure out what exactly is causing the issue. Not that 25% is necessarily unusual by any means depending on what settings you have enabled etc. But it seems to me in your case something is maxing out a core as it "continues to play" which is on another thread while the UI thread is hung. So after it finished you can't do anything.

MC is multithreaded and doesn't often do blocking work on the UI thread (there are some exceptions to this). So when something hangs, it is usually due to a call to the OS that the UI is waiting on. Something that normally isn't blocking, but is in the case of this particular system. In most cases I've seen it happens to be drivers or hardware issues. So anything you can provide that might point to the culprit would help.

How much more narrowing is needed? I can give more info if necessary, but as stated, the system is as close to a vanilla install as possible (it is a regular xeon og a regular server mobo, nothing extra except the soundcards). At the moment there is basically no other programs than MC installed. (the single other program installed is plextools, the problem was there before that, and it is not even running), the OS is out of the box. And when reading around in the forums, it seems like having some issues with the server-client setup is not unique. I am not saying this has to be a MC-problem, but I think it warrants some more checking out then "something other than MC is the problem"

And 25% for just playing on a modern quad-core should be unusual almost no matter what.

[flac.rules]

THis is a screenshot from the "analyse wait chain"-function in the task manager. However it should be noted that there are 6-7 threads in that list no matter what, even just after starting the program, and nothing playing.

[gvanbrunt]

Ok, Lets clear things up a bit. People here want to help you. They have explained the inner workings of MC and the underlying OS in terms of hangs etc to help sort that out. We have also pointed out certain facts (such as %cpu used) that you keep disagreeing with. That is your prerogative. However we have stated some facts for you to help to you sort out the "red herrings" - having to keep switching to explaining why doesn't fix your issue. You'll note that I said "does not necessarily" mean something unusual which means it may or may not be a symptom. So from here on out can we please keep to relaying relevant facts to us instead of arguing the validity of what's been said? It will make things go much faster and most likely will result in more people willing to help with your problem. We are trying to help you, not "deflect blame" from MC. Keep an open mind about what we are saying and where to look for the problem and you'll be much more likely to find the answer.

That said, lets also start from square one - things are too spread out over too many messages and threads. Can you please describe your setup (including brands) of the hardware in question, MC's setup on both computers, the network involved, and exactly how you reproduce the issue. I'm a little unclear about if this is happening on the client or the server for example.

[flac.rules]

Ok, Lets clear things up a bit. People here want to help you. They have explained the inner workings of MC and the underlying OS in terms of hangs etc to help sort that out. We have also pointed out certain facts (such as %cpu used) that you keep disagreeing with. That is your prerogative. However we have stated some facts for you to help to you sort out the "red herrings" - having to keep switching to explaining why doesn't fix your issue. You'll note that I said "does not necessarily" mean something unusual which means it may or may not be a symptom. So from here on out can we please keep to relaying relevant facts to us instead of arguing the validity of what's been said? It will make things go much faster and most likely will result in more people willing to help with your problem. We are trying to help you, not "deflect blame" from MC. Keep an open mind about what we are saying and where to look for the problem and you'll be much more likely to find the answer.

That said, lets also start from square one - things are too spread out over too many messages and threads. Can you please describe your setup (including brands) of the hardware in question, MC's setup on both computers, the network involved, and exactly how you reproduce the issue. I'm a little unclear about if this is happening on the client or the server for example.

Lets, it is nice people are willing to help, but it does from my perspective look like people are trying to deflect blame. Why? because of the fact that "the task manager might be showing the wrong info", "something else might be causing the high CPU", is nice info and all, but as long as these statements are just stated without any follow-up questions, It doesn't (on the surface) look like it is in the interest of finding the root cause. Furthermore, I haven't "kept disagreeing" with much at all, and I have continuously answered all questions asked, and tried to do relevant research and post more info.

I am sure you are trying to help me, and that is appreciated, look the above section as an explanation of how the thread so far has looked from my perspective, that might explain my reaction, not an attack .I know misunderstandings can occur when trying to put things in writing, and I am sure I also could be better when it comes to this, and I actually do appreciate the help.

The setup is the following

Server A: Has a MC 20 library server running, no AV, firewall open, 500 MB/s disk-array and gigabit net. It has  Xenon CPU and an Intel C224 Server board
Server B: Has a MC 20 client, connected to server A, no A, firewall open, gigabit net, 4 sound-cards which is sent to my speakers in different rooms in the house. It has a Xenon CPU, and a intel C224 server board

Both have Windows server 2012 R2 and uses the onboad NICs. Both have very little installed in terms of other programs, however server A also works as a regular file-server.

I don't remember the exact name of the hardware, I am going to check it out when i get back home.

The problem is on Server B, that is the client.

The problem seems to pop up any time I play anything from the library server (Server A), for some time (a couple of songs maybe). The CPU usage pops up to around 25% (pretty evenly distributed according to task manager). This does not happen if i play local files, but the high CPU load can continue for some time if it starts because of playing files from the library server, and I then switch over to local files. Which soundcard that is used seems not to be relevant (I have tried with two, I can check the other two) Neither does which of the two network ports are used. Please note the files played from the library server is NOT mapped up with file sharing on the Client. Server A is running a domain, which other computers on the network are joined too.

Let me know if any other info can be interesting, seeing as the setup is software-wise pretty simple, but still has a lot of "redundancy" in hardware, it should be pretty easy to check out a few things.

A final thing, both of the computer are almost always running remote desktop, as this is how I control them.

[mwillems]

Two thoughts/suggestions:

1) Do you have "play files from memory" enabled under Audio?  If so, try disabling it.  In my experience, it can lead to suboptimal results when used over a network (it's given me hangs and CPU spikes in the past, and network playback works much better for me when it's disabled).

2) Is it possible that MC is analyzing audio or building thumbnails in the background?  MC can take a while to analyze files in the background, and while that's going on I see 25% CPU usage until all files are analyzed.  With a large library that process could take days (it took me about a day and a half to two days to reanalyze about 60K files), but it only runs while MC is on and happens in the background. If you look under reporter you can see if either process is running in the background and whether your library is fully analyzed/thumbnails are fully built.  One thing to note is that while audio analysis only generally needs to happen once per file, thumbnail building happens on each MC instance for each file.  I've never seen either one of those hang the UI, mind you, but both are a possible cause of elevated CPU levels even on a client.  

These may not be your issues, but both are reasons I've seen elevated CPU usage for prolonged periods in the past, and play files from memory has definitely given me hangs.

P.S.- Also, have you made sure to set your client options and file access so that there's no transcoding happening?  That shouldn't necessarily create load on your client, but can make a server/client setup more sensitive to connectivity issues, so is worth checking just to rule things out.

[gvanbrunt]

Lets, it is nice people are willing to help, but it does from my perspective look like people are trying to deflect blame.

Fair enough, I suspected that was what you might be thinking which is why I said what I did. I can understand how it could seem like that to you. However keep this in mind: many of the people responding to this thread have successfully fixed issues like this in the past, and the cause was almost always not MC. Note that I said not always. There have been a few times where certain things were changed after finding the underlying cause. Some issues were caused by MC, and for some MC could be improved by changing things. But before that ever happened they ran down every other possibility first and narrowed things down considerably. To fix an issue we have to find the exact reproducible cause. Since this is not happening to everyone, we have to figure it out what is causing it in your particular case. Also consider that most people responding don't work for JRiver. Their only concern is seeing another user happy.

So enough about that. It seems to me that this is network related then. If local playback works fine. I have a couple of questions:

Can you give some detailed information about your network? IE the hardware involved, subnets used, what else might be on the network? Also what is going on on the server when a hang happens? Anything unusual there?

[flac.rules]

Two thoughts/suggestions:

1) Do you have "play files from memory" enabled under Audio?  If so, try disabling it.  In my experience, it can lead to suboptimal results when used over a network (it's given me hangs and CPU spikes in the past, and network playback works much better for me when it's disabled).

2) Is it possible that MC is analyzing audio or building thumbnails in the background?  MC can take a while to analyze files in the background, and while that's going on I see 25% CPU usage until all files are analyzed.  With a large library that process could take days (it took me about a day and a half to two days to reanalyze about 60K files), but it only runs while MC is on and happens in the background. If you look under reporter you can see if either process is running in the background and whether your library is fully analyzed/thumbnails are fully built.  One thing to note is that while audio analysis only generally needs to happen once per file, thumbnail building happens on each MC instance for each file.  I've never seen either one of those hang the UI, mind you, but both are a possible cause of elevated CPU levels even on a client.  

These may not be your issues, but both are reasons I've seen elevated CPU usage for prolonged periods in the past, and play files from memory has definitely given me hangs.

P.S.- Also, have you made sure to set your client options and file access so that there's no transcoding happening?  That shouldn't necessarily create load on your client, but can make a server/client setup more sensitive to connectivity issues, so is worth checking just to rule things out.

1) Play from memory is not enabled.

2) I don't think thats the problem(?) it has been several weeks since i imported audio, and I think the analyzing is happening on the server anyhow?

PS: I think so, i have set no transcoding under the media network-client options, the file access settings you are talking about, where are they?

[mwillems]

1) Play from memory is not enabled.

2) I don't think thats the problem(?) it has been several weeks since i imported audio, and I think the analyzing is happening on the server anyhow?

Analyzing can happen either place, it just depends on the auto-import settings.  One thing to try is to turn off auto-import on the client (it's not needed on a pure client as clients can't import anyway, and can cause trouble sometimes).

Again, the easy way to rule this out is to look under "Reporter" in the "Services & Plugins" menu tree.  It will show any background processes running.

PS: I think so, i have set no transcoding under the media network-client options, the file access settings you are talking about, where are they?

You want to make sure that your client PC has direct access to the actual media files on the server and that the filepath to the files on the client is the same as on the server.  For example, if your files are in a directory called X:/Audio on the server, you want to a) share that drive or directory over the network and then b) either map the share to drive X: on the client so the relative path is the same, or use UNC paths in the library database.

Then, in MC, under media network/client options make sure the option to "play a local copy" is checked.

If you don't do this, then the MC library server instance is relied on to serve and/or transcode files at all times, which can lead to suboptimal network results.  If your client has direct access to the files, it just plays the files directly; in that case the library server is serving only the library, not the media files themselves.

EDIT: reading above it looks like you aren't using UNC paths or mapped drives on this specific computer; it will probably help the situation if you do.

[flac.rules]

Fair enough, I suspected that was what you might be thinking which is why I said what I did. I can understand how it could seem like that to you. However keep this in mind: many of the people responding to this thread have successfully fixed issues like this in the past, and the cause was almost always not MC. Note that I said not always. There have been a few times where certain things were changed after finding the underlying cause. Some issues were caused by MC, and for some MC could be improved by changing things. But before that ever happened they ran down every other possibility first and narrowed things down considerably. To fix an issue we have to find the exact reproducible cause. Since this is not happening to everyone, we have to figure it out what is causing it in your particular case. Also consider that most people responding don't work for JRiver. Their only concern is seeing another user happy.

So enough about that. It seems to me that this is network related then. If local playback works fine. I have a couple of questions:

Can you give some detailed information about your network? IE the hardware involved, subnets used, what else might be on the network? Also what is going on on the server when a hang happens? Anything unusual there?

My network only has one subnet, I have a HP 1810 managed gigabit switch, both servers are directly connected to this switch. I have a RT66U router for network access, but traffic should go directly through the switch with my setup. I have 3 other computers connected to the media server. I don't have the same problem on them, but they have the drives mapped up.

The server seems to be chugging along at around 0% load, so it seems to make no difference server-side.

Could it be some kind of domain-related problem? Server 2012 R2 forces computers connected to it to join a domain.

[Matt]

If you're having an MC hang, please use Logging and post the results here.  Thanks.

[gvanbrunt]

Could it be some kind of domain-related problem? Server 2012 R2 forces computers connected to it to join a domain.

Domain shouldn't really affect things network wise. However using server OS's isn't of the norm from MC. I know some users had issues with Windows Home Server in the past. I think those issues were all resolved or were related to some oddball things only specific to WHS.

[flac.rules]

If you're having an MC hang, please use Logging and post the results here.  Thanks.

To clarify, it doesn't hang permantly, it "springs to life", after a minute or two. I don't know how this affects logging, but i will try to log and see what it gives.

[gvanbrunt]

try to keep track of the times when it "hangs" then.

Crazy question but all your drivers especially network card drivers are the latest available from the manufacturer of the part? For example even if it's a Dell computer you get the driver from ATI for video card etc?

[glynor]

To clarify, it doesn't hang permantly, it "springs to life", after a minute or two. I don't know how this affects logging, but i will try to log and see what it gives.

That's exactly the type of logs that can be helpful.

[glynor]

Fair enough, I suspected that was what you might be thinking which is why I said what I did. I can understand how it could seem like that to you. However keep this in mind: many of the people responding to this thread have successfully fixed issues like this in the past, and the cause was almost always not MC.

Absolutely everything gvanbrunt said in this thread gets a +1 from me.

I do not see this behavior.  I have seen this behavior on some systems in the past, and have resolved it.  For me, personally, it was either:

* Anti-Virus
* Drivers (in my case, network drivers and disk drivers)

Computers are a complex set of code stacked on one another and the software at the "top" often gets the "blame" for things going on underneath.  That doesn't mean it couldn't be MC's fault.  It certainly can be, and could be here, but we don't know.  All we do know is that we don't see it, and JRiver doesn't see it, and it is IMPOSSIBLE to diagnose a fault or bug (especially hangs, which are the worst of all problems to try to fix) if you can't reproduce it.

We NEED TO figure out why it happens to you, and not to us.  Things that we ask you to do or check that seem like they are deflecting are us trying to figure out what makes your system special.

That, and running Windows Server is absurdly complex.  If you are running it, and you aren't an IT person, you are probably in way over your head.  It is really designed to be run by people with years of training and ongoing support contracts.  I can't speak for JRiver, but it is almost certainly an unsupported configuration for MC.

[flac.rules]

If you're having an MC hang, please use Logging and post the results here.  Thanks.

Attached is the log, I will post  a quick rundown of what happend.

I start the program
Zone "kjøkken" i chosen, and the playlist "Fest3" is started to play in that zone.
I skip a few songs to provoke the problem to happen as fast as possible
The song "a-ha - analogue" is then played in its entirety

1.15 out in the song CPU load jumps up to 27% (local time 08.31.22)
The UI hangs, I notice this at local time 08.33.41, however the song timer is on 03.06, indicating the actual event to be at 08.33.13
At the end of the song it uses a long time to jump to the next song.
The next song is started at local time 08.35.09 (about a minute "too late")
I stop the song, and exit the program, I exit at local time 08.35.29
At 08.38.19 (around, I didn't catcht it at the exact second), the MC-process avtually ends.

Hope this can be useful.

[flac.rules]

You want to make sure that your client PC has direct access to the actual media files on the server and that the filepath to the files on the client is the same as on the server.  For example, if your files are in a directory called X:/Audio on the server, you want to a) share that drive or directory over the network and then b) either map the share to drive X: on the client so the relative path is the same, or use UNC paths in the library database.

Then, in MC, under media network/client options make sure the option to "play a local copy" is checked.

If you don't do this, then the MC library server instance is relied on to serve and/or transcode files at all times, which can lead to suboptimal network results.  If your client has direct access to the files, it just plays the files directly; in that case the library server is serving only the library, not the media files themselves.

EDIT: reading above it looks like you aren't using UNC paths or mapped drives on this specific computer; it will probably help the situation if you do.

I have checked this before, I didn't have the problem then (or so I thought), however I do not wish to map up the drives, I wish to use the "pure" Server-client function in MC.

Anyhow, I checked it again, and (this time) was able to reproduce the problem even when the drive was mapped up and "Use local files" is checked. I have several other computers connected to the library server with that setup not having this problem, so it does indicate something is happening in that specific computer.

[flac.rules]

Would it be helpful if i install a (virtual) version of for instance windows 7 on the same computer (server B), and see if that exhibits the same problems? (it will take some time to setup)

[JimH]

Please see this thread:
http://yabb.jriver.com/interact/index.php?topic=93174.0

[gvanbrunt]

Just a reminder to check your network drivers. I can tell you from experience drivers for server class network cards tend to have issues. They are like graphics card drivers where manufacturers go for performance over stability. I've had to mess with drivers and their settings (yes the actual drivers have settings) to get things working correctly on a few servers I've maintained over the years. To give you an example, I once built a new Exchange mail server on some very high end server hardware. It came up and worked correctly. Then all of a sudden it died a couple of hours later and would no longer let clients connect. This went on for about 36 hours of which I had no sleep and resulted in a support call to MS that was kicked all the way up to the developers of the software itself. We finally found the issue with the network card and Intel changed the driver as a result. Keep in mind that everything else on that server worked perfectly without network issues. Also keep in mind that "no one else" had seen the issue before with the same network card and Exchange. It was my particular set-up that killed things.

And what Glynor says about server OS's is true. While they look similar to their workstation counterparts, there are a LOT of different things happening below the surface. They are tuned differently and that includes graphics capabilities. They are not meant to run graphic intensive applications and tuned to support disk and network IO instead. Even as a server MC can rely on graphics for transcoding etc. I never wrote the software so I can't tell you how often it gets used and where. Another place would be during autoimport where it analyses video etc. Neither of us is saying it won't work on a server OS, just that it isn't optimal.

My bet is on a server category Nic doing something odd. Which is why I asked about the drivers. I need to know the make and model and the driver version as well.

[kstuart]

Is your browsing engine (in JRiver MC Options) set to Chromium or Internet Explorer ?

[6233638]

Just a reminder to check your network drivers.
[…]
My bet is on a server category Nic doing something odd. Which is why I asked about the drivers. I need to know the make and model and the driver version as well.

Definitely worth checking.
While it's not a server-grade NIC in my system, the latest Intel drivers on Windows Update for mine introduce significant DPC latency problems. (more than 6000µs)
 
I had to revert to the stock Windows 8.1 driver and block the update.

[JimH]

At the moment there is basically no other programs than MC installed. (the single other program installed is plextools, the problem was there before that, and it is not even running), the OS is out of the box.

Could you try uninstalling plextools?

[glynor]

Attached is the log, I will post  a quick rundown of what happend.

The log you provided wasn't collected using the log file collection tool.

Please recreate the issue and follow the instructions here to collect the logs:
http://wiki.jriver.com/index.php/Logging

It would also be helpful to re-iterate any points you already addressed in the long thread above with bullets (to make it so everyone doesn't need to dig through the thread looking for things already asked and answered).  As in:

* Anti-virus status
* Primary settings under Options > Video and Options > Audio (screenshots are handy for this)
* Output Format settings in the DSP
* The other stuff discussed here: http://wiki.jriver.com/index.php/Troubleshooting_Guide#Look_For_Patterns
* Brief rundown on what steps you've taken to date to address this issue (so we can rule things in or out)

You don't have to go into detail (we can get that above for things you've already explained) but just quick bullets covering the major points.

[flac.rules]

The log you provided wasn't collected using the log file collection tool.

Please recreate the issue and follow the instructions here to collect the logs:
http://wiki.jriver.com/index.php/Logging

It would also be helpful to re-iterate any points you already addressed in the long thread above with bullets (to make it so everyone doesn't need to dig through the thread looking for things already asked and answered).  As in:

* Anti-virus status
* Primary settings under Options > Video and Options > Audio (screenshots are handy for this)
* Output Format settings in the DSP
* The other stuff discussed here: http://wiki.jriver.com/index.php/Troubleshooting_Guide#Look_For_Patterns
* Brief rundown on what steps you've taken to date to address this issue (so we can rule things in or out)

You don't have to go into detail (we can get that above for things you've already explained) but just quick bullets covering the major points.

Can you specify a bit more what is wrong with the log? I followed in the wiki (or so I thought) to enable logging on the same page you linked too, and then sent over the log created. Did I send the wrong log or something?

Sorry I haven't been able to try the most recent suggestions yet, I have been on some work travel, I will look at it this evening.