Spinning circle when STOPPING playback

[BobPhoenix]

Originally Posted by sflamm

That should be always - why would the video not go to shared storage independent of any VM?

Sounds like you are having a serious issue.

Nope no real problems although I would prefer a NON port multiplier solution for my recording drives. When the recording finishes I check it to make sure it is OK then transfer it to my unRAID VM.

I have multiple recording drives because I don't always get them moved over before the recording drives fill up. Also I can spread around the recordings to different drives so that I don't have bad recordings due to overloaded drives. 8 recordings to one drive at a time would be a bit much and I just don't trust drives anymore.

I don't want to record to a RAID array because if I loose more drives than are protected by parity I don't loose everything with independent drives. I've lost too many recordings already from the hardware RAID cards I use to use (3Ware 9500S-12's). Drives got dropped from my 3Ware controllers or went bad too often and I would loose everything on the array. I don't trust any raid scheme that doesn't have drives with independent file systems any more.

I like unRAID because it is very easy to setup. It gives parity protection for one drive failure. It spins down drives that are not being accessed so I use less electricity since 95% of the time they are spun down. Most importantly if I have drive failures I can get back everything off my array that wasn't on a failed drive. I.E. If I have two drives die at once or a second drive failure while rebuilding a failed drive I don't loose all of my data - only that data on the failed drives. I can even get back data on failed drives with consumer level/free file recovery programs. I don't need to use programs that work with RAID arrays since each drive has an independent file system. The recovery program does have to understand ReiserFS but many do and the help on the unRAID forums make up for any problems.

I don't want to record to my unRAID array because the write performance is not real good compared to traditional RAID. It takes 4 IO operations for a single write if writing directly to the array. Or if using a cache drive you only have a single drive that gets written to at a time. I can batch up the moves to all be over night when I'm not recording anything or at least not to the drive SageTV is recording to anyway.

Make sense now?

[Skirge01]

@gplasky: I am running DX11 on my Windows 7 client and DX9.0c on the XP service VM.

Originally Posted by sflamm

There is a UI to maintain it - runs from a Windows Desktop machine (see installation instructions).

Looks like it needs to be run from either a Windows Server 2012 or Windows 8 OS. Am I reading that right?

It does not support PCI passthrough - but why would you need that? Setup all your video sources as network encoders - more flexible in the long run.

I use pass-through for the USB card in the PCI slot, which has the HDPVRs. I'm a little confused on the network encoder option, but I do recall seeing a thread about that on here recently, so I'll check that out. I also pass through my LSI M1015 card to the Windows 7 OS because vSphere still doesn't support 3TB drives with 3TB VMDKs; instead, I just pass the entire controller to the guest OS. If Hyper-V doesn't have that limitation, then it obviously wouldn't matter.

I understand it is work to try given that VMWare is installed. But unless you can get the VSphere install working correctly it might be a good alternative.

Given your setup I am wondering why not just run W2012, install SageTV server and use it's built in storage services? No VMs and should run like a champ.

(Emphasis added.) Yes, but so should vmware. I could very well switch over and still have an issue; the same, or even new one(s). I'd also need to learn a whole new OS. Just to clarify, when you write "W2012", you mean the free Hyper-V Server 2012, correct? I ask because you said no VM's, but I'd still need to run pfSense in a VM. Also, one of the reasons I separated the OS's into VM's was so that I could take one down without impacting the others.

[gplasky]

Originally Posted by Skirge01

@gplasky: I am running DX11 on my Windows 7 client and DX9.0c on the XP service VM.

I have the latest DX9 installed on my Win 7 machine. I had a couple of apps that needed it. (MPC-HC being one of them.) It won't hurt to install it. But I'm beginning to think your issue is more of a network issue. Have you tuned any of the virtual NIC settings?

Do you have any extenders and do they act the same?

Gerry

[Skirge01]

Originally Posted by gplasky

I have the latest DX9 installed on my Win 7 machine. I had a couple of apps that needed it. (MPC-HC being one of them.) It won't hurt to install it. But I'm beginning to think your issue is more of a network issue. Have you tuned any of the virtual NIC settings?

Do you have any extenders and do they act the same?

Gerry

Would I be replacing DX11 with DX9 or can they be installed side-by-side? I did a very quick Google search and didn't see anything.

I have not tuned any of my NIC settings. They're all at default settings. I am running the vmxnet3 NIC on all VMs, which is the newest one. I could revert to the older vmxnet2 and give that a shot. If there are any specific settings you have in mind, let me know.

I have no extenders, just this one client PC.

UPDATE: Went ahead and installed DX9 without issue. I then tried both EVR and VMR9 in the Sage client without any difference.

[sflamm]

I don't want to record to a RAID array because if I loose more drives than are protected by parity I don't loose everything with independent drives. I've lost too many recordings already from the hardware RAID cards I use to use (3Ware 9500S-12's). Drives got dropped from my 3Ware controllers or went bad too often and I would loose everything on the array. I don't trust any raid scheme that doesn't have drives with independent file systems any more.

Sorry to hear that you had problems - the very likely reason is you used drives (non-enterprise) that did not support TLER which will cause drives to drop out of the array. It is important to use Enterprise drives with real hardware RAID.

Your concern about reliability of RAID in general is unfounded. Quite the opposite they are extraordinarily reliable and have outstanding performance. I been running multiple RAID 5 arrays for years with not a single problem.

It spins down drives that are not being accessed so I use less electricity since 95% of the time they are spun down

All hardware RAID controller can be configured to do this as well.

If I have two drives die at once or a second drive failure while rebuilding a failed drive I don't loose all of my data

Redundancy depends on how you configure your RAID. Raid 5 (1 drive), Raid 6 (2 drives) etc. You can also use a hot spare to auto rebuild on a drive failure.

I don't want to record to my unRAID array because the write performance is not real good compared to traditional RAID

Exactly - which is why you should avoid all this unnecessary complications of writing, verifying and copying and simply go back to using a high performance RAID5 array.

It can all be easily solved by fixing the problem you have with your RAID by using the appropriate drives.

[Skirge01]

I played around with some NIC settings (larger, matching buffers, disabling VLAN priorities) with no change. I also tried changing the computers to focus on background services with no change. I also tried changing the write buffer settings on the individual shared drives with no change. I'm considering turning off write caching on them entirely just to see what happens. I'm also going to try the older vmxnet2 NIC in the VMs and possibly the e1000. I'd try jumbo frames, but my client PC's NIC doesn't support it.

[sflamm]

Looks like it needs to be run from either a Windows Server 2012 or Windows 8 OS. Am I reading that right?

I believe that is correct

I use pass-through for the USB card in the PCI slot, which has the HDPVRs. I'm a little confused on the network encoder option

Network Encoder allows the video sources to be physically located elsewhere (not in the same box)

I also pass through my LSI M1015 card to the Windows 7 OS because vSphere still doesn't support 3TB drives with 3TB VMDKs; instead, I just pass the entire controller to the guest OS. If Hyper-V doesn't have that limitation, then it obviously wouldn't matter.

I believe that Hyper-V allows storage cards to be assigned to particular VMs (that is one of the pass through items it does do)

Just to clarify, when you write "W2012", you mean the free Hyper-V Server 2012, correct?

I actually meant using Windows 2012 not Hyper-V (to simplify everything)

I'd still need to run pfSense in a VM

If you still want to run pfSense in a VM you can do that with W2012 as it has the Hyper-V role built in.

[Skirge01]

I'm not sure I want to put out $400 when I'm not certain it'll resolve the issue.

[razrsharpe]

Originally Posted by gplasky

Wed 1/2 22:22:02.061 [AWT-EventQueue-0@128a415] Get Imported TV Files - Done
Wed 1/2 22:22:02.061 [AWT-EventQueue-0@128a415] Get Recorded Files - Start
Wed 1/2 22:22:02.092 [AWT-EventQueue-0@128a415] Get Recorded Files - Done
Wed 1/2 22:22:02.093 [AWT-EventQueue-0@128a415] Get Scheduled Files - Start
Wed 1/2 22:22:02.823 [AWTThreadWatcher-SAGETV_PROCESS_LOCAL_UI@374534]  EventThread-SAGETV_PROCESS_LOCAL_UI Hang Detected - hang time = 750
... 
Wed 1/2 22:22:11.072 [AWTThreadWatcher-SAGETV_PROCESS_LOCAL_UI@374534]  EventThread-SAGETV_PROCESS_LOCAL_UI Hang Detected - hang time = 8999
Wed 1/2 22:22:11.500 [AWT-EventQueue-0@128a415] Get Scheduled Files - Done
Wed 1/2 22:22:11.500 [AWT-EventQueue-0@128a415] Start applying filters
Wed 1/2 22:22:11.501 [AWT-EventQueue-0@128a415] Done applying filters
Wed 1/2 22:22:11.501 [AWT-EventQueue-0@128a415] Start Grouping
Wed 1/2 22:22:11.539 [AWT-EventQueue-0@128a415] Done Grouping
Wed 1/2 22:22:11.539 [AWT-EventQueue-0@128a415] Starting EpSorting

Originally Posted by Skirge01

Got it. Thanks, Gerry. I'll check the DirectX status when I get home. Do you know if that data (i.e. scheduled and episode list) is stored on the client or the server? Could this be a wiz.bin corruption issue?

That is all TVE logging information. that code runs once when TVE is first loaded (ie entered from the main menu) or reloaded (ie when playback is ended and returning to TVE) and is responsible for generating the list of media files to display. That is a really really long time... i never see wait times like that (or at all) when coming back (or entering) TVE. I might try temporarily disabling TVE (or using the default recording screen to start playback) to see if that improves it... then at least we know where the problem is

EDIT: Also is the problem stopping all types of playback or just TV (from TVE)?

[sflamm]

I'm not sure I want to put out $400 when I'm not certain it'll resolve the issue.

Understandable. However, that is a known common problem with a solution.

http://www.wdc.com/en/library/other/2579-001098.pdf

http://en.wikipedia.org/wiki/Error_recovery_control

[BobPhoenix]

Originally Posted by sflamm

Sorry to hear that you had problems - the very likely reason is you used drives (non-enterprise) that did not support TLER which will cause drives to drop out of the array. It is important to use Enterprise drives with real hardware RAID.

I did use enterprise drives on an early array but they are too expensive and I had a 25% failure rate on them (2 out of 8) which is higher than the failure rate of my non enterprise drives - purchased 100s. Lately I've had bad luck with the WD Green's - got 8 DOA in a row. Which is why I'm buying WD Red's now. Plus the Red's have longer warranties. But I wouldn't consider them enterprise drives even though they are intended for NAS boxes.

Originally Posted by sflamm

Your concern about reliability of RAID in general is unfounded. Quite the opposite they are extraordinarily reliable and have outstanding performance. I been running multiple RAID 5 arrays for years with not a single problem.

That is good. I still prefer drives with independent file systems - data recovery is easier. Never lost a byte on my unRAID servers.

Originally Posted by sflamm

All hardware RAID controller can be configured to do this as well.

I'm sure that is true with all modern controllers but I don't think it is true with my 3Ware 7506-12's and 9500s-12's.

Originally Posted by sflamm

Redundancy depends on how you configure your RAID. Raid 5 (1 drive), Raid 6 (2 drives) etc. You can also use a hot spare to auto rebuild on a drive failure.

A hot spare won't do any good on Raid 5 if another drive gets droped while rebuilding on the hot spare. I had that happen and lost an array of data that way. Note never owned a RAID6 capable controller. Spent enough on my 9500s's and 7506's didn't want to re-spend it again.

Exactly - which is why you should avoid all this unnecessary complications of writing, verifying and copying and simply go back to using a high performance RAID5 array.

It IS necessary because AS I SAID I will not go back to a system that doesn't have independent file systems on each disk. I could loose 90% of my drives and still get some of my data back. And while the following is completely preventable with unRAID (or a hardware controller probably) it is another reason I insist on independent file systems: I burned up 12 drives at the same time when a case fan failed on a parity check and the drives over heated. The other 10 drives in 5in3 cages were fine. This happened in the last year. If I had configured unRAID to power down for drive overheat (like you can probably do with a hardware controller) I wouldn't have had to replace any drives. I can say that even though the drives were overheated and I can't trust them anymore I was able to recover all the data on them. My 9500s-12s would have burned them up and dropped more than 1 drive from the array and lost ALL data.

It can all be easily solved by fixing the problem you have with your RAID by using the appropriate drives.

Enterprise drives are good but they won't fix everything.

[sflamm]

It IS necessary because AS I SAID I will not go back to a system that doesn't have independent file systems on each disk. I could loose 90% of my drives and still get some of my data back.

Suite yourself but doesnt make a lot of sense. All real file servers use hardware raid and without data loss. You are going to a tremendous amount of trouble to avoid the obvious solution. The odds of multiple drive failure is super-low and if you run RAID 6 almost 0.

The odds of you losing data with your individual drives running is much much higher.

Yes Enterprise drives cost more and that's for a reason.

A hot spare won't do any good on Raid 5 if another drive gets droped while rebuilding on the hot spare. I had that happen and lost an array of data that way.

Again - address the real problem which is the drives dropping out. This is not a problem anyone else has because they use the right drives.

If I had configured unRAID to power down for drive overheat (like you can probably do with a hardware controller) I wouldn't have had to replace any drives.

Enterprise drives are designed to run continously drives and do not overheat. That is unheard of...

If you are ever interested buy some of these (WD RE series):

http://www.newegg.com/Product/Produc...82E16822236352

[razrsharpe]

Originally Posted by razrsharpe

That is all TVE logging information. that code runs once when TVE is first loaded (ie entered from the main menu) or reloaded (ie when playback is ended and returning to TVE) and is responsible for generating the list of media files to display. That is a really really long time... i never see wait times like that (or at all) when coming back (or entering) TVE. I might try temporarily disabling TVE (or using the default recording screen to start playback) to see if that improves it... then at least we know where the problem is

EDIT: Also is the problem stopping all types of playback or just TV (from TVE)?

I was able to somewhat reproduce the UI hang on stopping playback and returning to TVE. It is only 750ms on my test setup, so not very noticeable, but it is in the log. I just uploaded a new version of TVE to the repository that only generates the list of scheduled recordings if you browse them (ie it will skip generating the scheduled recording list upon returning to TVE after playback stopped). What is interesting is that during normal operation (ie entering TVE from the main menu) it only takes 1-2ms to generate the list of scheduled recordings

[BobPhoenix]

Originally Posted by sflamm

Suite yourself but doesnt make a lot of sense. All real file servers use hardware raid and without data loss. You are going to a tremendous amount of trouble to avoid the obvious solution. The odds of multiple drive failure is super-low and if you run RAID 6 almost 0.

The odds of you losing data with your individual drives running is much much higher.

I believe in luck not odds. Mine appears to be all bad. No matter how good the parity protection I will loose at least one more drive than it covers.

Originally Posted by sflamm

Yes Enterprise drives cost more and that's for a reason.

I agree they should last longer but I can get 2 sometimes 3 times the storage for the same cost.

Originally Posted by sflamm

Enterprise drives are designed to run continously drives and do not overheat. That is unheard of...

Just checked the WD site and the Operating Temperature specs are better for WD Green drives (60C) as opposed to the SATA RE drives (55C) - 3TB size. My drives got to 60+C during the parity check. The electrical specs show when operating they are not even close but I wouldn't expect it to be since the Greens are designed to be as low power as possible. The specs for Green list standby/sleep but can't tell what it would be for the RE drives. IDLE power is 5.5 watts for Green - 8.6 watts for RE. I expect standby/sleep (what a spun down drive should draw) to be closer but don't know for sure. The RE drive I checked was this. The green drive checked was this.

Originally Posted by sflamm

If you are ever interested buy some of these (WD RE series):

http://www.newegg.com/Product/Produc...82E16822236352

That was the 2TB model of the 3TB I got specs for.

[Skirge01]

Originally Posted by razrsharpe

I was able to somewhat reproduce the UI hang on stopping playback and returning to TVE. It is only 750ms on my test setup, so not very noticeable, but it is in the log. I just uploaded a new version of TVE to the repository that only generates the list of scheduled recordings if you browse them (ie it will skip generating the scheduled recording list upon returning to TVE after playback stopped). What is interesting is that during normal operation (ie entering TVE from the main menu) it only takes 1-2ms to generate the list of scheduled recordings

Thanks for the attempt, Razr. I'll use this one for a little bit and report back. Out of curiosity, what happens when the scheduled recordings list is generated in TVE? Is this an existing list that Sage already has or does it update it on the spot, based on favorites? I ask because my upcoming recordings list usually has about 20 shows listed, but I happen to have (what I consider is) an unusually high number of favorites which hovers around 100, consisting of a large number of movies I'm waiting to reappear in the EPG, many with channels specified.

[razrsharpe]

Originally Posted by Skirge01

Thanks for the attempt, Razr. I'll use this one for a little bit and report back.

on my test setup it is marginally faster coming back to TVE... hopefully it works for you too

Out of curiosity, what happens when the scheduled recordings list is generated in TVE? Is this an existing list that Sage already has or does it update it on the spot, based on favorites?

The list of scheduled recordings is generated by calling the sage api function GetScheduledRecordings(). It should be the same list (grouped by show of course) that Sage displays on its recording list, and that the webserver shows on its page. It is a list of everything that is currently scheduled to record.

[Skirge01]

It is coming back faster, but not always. I played with a bunch of settings, but nothing has changed. I reinstalled Gemstone and it's doing it there, too. My next step is to reinstall the original HDD and see if that fixes the delay; perhaps something is awry with the SSD.

I noticed something else with TVE, but I'll post that bug in the TVE thread.

[sflamm]

did you install and run the Intell SSD toolbox?

http://downloadcenter.intel.com/Deta...&DwnldID=18455

[razrsharpe]

Originally Posted by Skirge01

It is coming back faster, but not always. I played with a bunch of settings, but nothing has changed. I reinstalled Gemstone and it's doing it there, too. My next step is to reinstall the original HDD and see if that fixes the delay; perhaps something is awry with the SSD.

I noticed something else with TVE, but I'll post that bug in the TVE thread.

can you post the log? is it pausing the UI HANG in the same place?

[Skirge01]

Originally Posted by sflamm

did you install and run the Intell SSD toolbox?

http://downloadcenter.intel.com/Deta...&DwnldID=18455

No, I'm not running an Intel SSD. But, I did just install it to see what it would do on another brand of SSD. It gave me a bunch of internal info on it, but I have no idea what I should be looking for. All options except for drive info and smart status are disabled.

Originally Posted by razrsharpe

can you post the log? is it pausing the UI HANG in the same place?

I restarted the client for another reason and now I can't get it to hang again. That's what happened yesterday, too. It would work fine when I was trying to create the hang and then when we were just watching TV, it would hang. I'll let it go for a while longer and see if it comes back. I'll keep you posted.