xbmc users, how are you dealing with spin ups?

[generalz]

Everything is working normal as far as unraid goes. But XBMC has been freaking out over the delay of the spin up and often wants to remove the media from the library causes problems and crashes when I pause something and leave and come back.

 

I've turned it up from 15minutes to 60 and that has resolved the pausing issues, I just stop it if I'm going to leave for longer.

 

One thing I have noticed is that the whole system pauses when a drive is spun up. The web interface, any downloads, backups or file transfers.

 

Does your box pause for a spin up?

 

Unraid setup:

ECS 885GM v2 /w AMD 945X4 /w 4Gb DDR3

7 Drives including Parity and Cache

Intel Gigabit PCI network card

Currently have the Bri10 PCIe SAS Card

 

Running (all installed to cache)

apcupsd

Sab, Sick, Couch, Transmission, Pyload

Mysql (for xbmc)

CrashPlan

SimpleFeatures package

Unmenu

 

 

3 xbmc setups (eden)

1 xbuntu machine (xbmc live)

2 windows machines

 

 

 

 

[PeterB]

I've just tried to provoke this problem - without success.  There seems to be enough data buffered somewhere for play to resume immediately.  By the time more data is required for display, the disk has spun up and play continues without a hitch.

 

My setup:

  Asus E45M1-I Deluxe (AMD E450 processor including Hudson chipset/Radeon HD 6320 Graphics)

  4GB DDR3 1333 RAM

  OpenELEC v1.95rc5

 

  unRAID server as listed in my sig.

 

  Playing through Sony Bravia KLV-46X450A

[Bizarroterl]

I tried the tag playlisttimeout in advancedsettings but it didn't seem to make a difference, though I really didn't give it much of a try.  I just got used to clicking on No to remove from library (when it appears) and hitting the play button a second time.

 

Either playlistretries or playlisttimeout should be where we can fix this.  Maybe someone is using it successfully. 

 

I suspect that the problem will be install specific.  IE if you have only 2 drives to spin up you'll see data much sooner than if you're spinning up 5.  I'm waiting for v5 and I'm going to start swapping out drives with 4TB ones.  That'll reduce the # of drives I have per share and reduce the number of drives I need to spin up.

[ljh89]

I use a Raspberry Pi with raspBMC for of xbmc, it seems fine with the spin up delay.

[Joe L.]

I use cache_dirs for nearly instant display of listings for XBMC.  It will make the biggest difference.

Once listed, and media subsequently selected to play, XBMC seems to wait just fine for a specific disk to spin up. 

[PeterB]

I suspect that the problem will be install specific.  IE if you have only 2 drives to spin up you'll see data much sooner than if you're spinning up 5

 

I find this unconvincing ... unless you have the five drives in the same spin-up group.  All the streaming data for the movie being paused/played should exist on a single drive.

[PeterB]

I use cache_dirs for nearly instant display of listings for XBMC.  It will make the biggest difference.

Once listed, and media subsequently selected to play, XBMC seems to wait just fine for a specific disk to spin up.

 

Joe, I think that you're only considering the initial play of a film, and I believe that works fine in all situations.

 

What the OP is referring to is the spinup delay which occurs if a film is paused, mid play, for so long that the data drive spins down.  The OP says that it creates a problem in his system. 

 

In my case, there is no problem.  Play resumes immediately using, I presume, buffered data.  I then hear a click from the drive as it spins up two to three seconds after I un-pause.  The spinup causes no perturbation in the play - it just carries on seamlessly.

 

Now, my question is:

Why do I not experience the problem which the OP is reporting?  Is there really a difference in buffering between the two environments, is he using a much higher stream, for which the buffers are insufficient, or do his drives take very much longer to spin up and come back on line?

[generalz]

after increasing the spin down time up to an hour that has fixed when its xbmc is paused for an extended period of time. If I'm gone for more then an hour I just hit stop and it creates a bookmark in mysql that I can use to resume later.

 

But my main problem now is that when starting a movie or tv show with all disks spun down it causes xbmc to crash 8 out of 10 times in windows and linux it either wants to remove the item from library or it starts playing and locks up/ crashes

 

 

now that you mention multiple drives spinup I'm wondering if the the thumbnails profile redirect is doing it.  I think I have it redirected to a user share instead of the cache drive. I'll have to check when i get home.

 

I was using cache dirs with 4.7  but with 5b14 and simple features when I go to shutdown or reboot it hangs with open files/shares even when shutting it off first. I have to kill it manually and I always forget to do that  so I stopped using it

[Bizarroterl]

I suspect that the problem will be install specific.  IE if you have only 2 drives to spin up you'll see data much sooner than if you're spinning up 5

 

I find this unconvincing ... unless you have the five drives in the same spin-up group.  All the streaming data for the movie being paused/played should exist on a single drive.

 

In my case I do have the drives in a spinup group.  My groups follow the shares, IE I have a HD Movies share and the drives in that share are all in a spinup group.  Once the drives spin up I have no further problems as I've set my period to and hour.  Of course, if I watch a movie and then go to watch a TV show I go through the same process as TV Shows are on a different share/spin up group.

 

 

I'm running XBMCBuntu.

[Bizarroterl]

I have a few XBMC boxes and a central database for them.  One of the common things to do with the central database is to also put all the thumbnails on a common share.  What I did on my array is I added a SSD to it.  It never spins down and is instantly available for thumbnails.  This has worked very well, both in v10 and v11. 

[generalz]

So does your Unraid system pause when waiting for one drive to spin up?

[Joe L.]

So does your Unraid system pause when waiting for one drive to spin up?

Every system is different., but most will if they have IDE controllers, or SATA controllers with firmware built on an IDE disk firmware base, the concept of master/slave disks still applies and only one command is active to a disk on a controller at a time.  If one of a pair is spinning up, the other must wait.  (apparently a hardware limitation) 

 

In addition, SMB communications from a Windows PC is apparently single threaded.  If one request is outstanding (waiting for a spin-up) another will wait.  (Apparently a Windows issue)

 

Both are why spin-up groups were added to unRAID.  That is supposed to prevent one spinup from blocking communications to the other disk sharing the communications channel.

 

[generalz]

huh, maybe i just need to skip the spin up for now and plan on coordinating spin up groups and split levels and rebuild when I gertsome new disks

[PeterB]

In addition, SMB communications from a Windows PC is apparently single threaded.  If one request is outstanding (waiting for a spin-up) another will wait.  (Apparently a Windows issue)

 

Ah, maybe that is the issue!  I am using NFS exclusively, and get no delays.

[generalz]

So does your Unraid system pause when waiting for one drive to spin up?

Every system is different., but most will if they have IDE controllers, or SATA controllers with firmware built on an IDE disk firmware base, the concept of master/slave disks still applies and only one command is active to a disk on a controller at a time.  If one of a pair is spinning up, the other must wait.  (apparently a hardware limitation) 

 

In addition, SMB communications from a Windows PC is apparently single threaded.  If one request is outstanding (waiting for a spin-up) another will wait.  (Apparently a Windows issue)

 

Both are why spin-up groups were added to unRAID.  That is supposed to prevent one spinup from blocking communications to the other disk sharing the communications channel.

 

 

 

Maybe I will shut off all that stuff in the bios. I will have to move my cache drive from the onboard controller to the sas.

 

Turns out my thumbs folder is not using a user share   So I will need to bust out a packet sniffer and see what xbmc is doing.

 

<advancedsettings>
    <videodatabase>
        <type>mysql</type>
        <host>192.168.1.111</host>
        <port>3306</port>
        <user>xbmc</user>
        <pass>xbmc</pass>
        <name>xbmc_video</name>
    </videodatabase> 

    <musicdatabase>
        <type>mysql</type>
        <host>192.168.1.111</host>
        <port>3306</port>
        <user>xbmc</user>
        <pass>xbmc</pass>
        <name>xbmc_music</name>
    </musicdatabase>

<pathsubstitution>
  <substitute>
    <from>special://masterprofile/Thumbnails/</from>
    <to>smb://192.168.1.111/cache/.custom/Thumbnails/</to>
  </substitute>
</pathsubstitution>

</advancedsettings>

[PeterB]

<advancedsettings>
    <videodatabase>
        <type>mysql</type>
        <host>192.168.1.111</host>
        <port>3306</port>
        <user>xbmc</user>
        ....

 

If you add your unRAID server to the hosts file on xbmc/OpenELEC, you can use the host name instead of having to use the quad-dotted address.

 

[generalz]

Yea I used to have it that way but sql was way slow. I would post the link to the xbmc forum but its down. anyway its about optimizing sql. one of them was turning off dns resolution.

 

 

# * Query Cache Configuration

#

query_cache_limit      = 1M

query_cache_size        = 16M

thread_concurrency = 4

skip-name-resolve

 

and there is the my.cnf

[PeterB]

Yea I used to have it that way but sql was way slow.

 

Really?  I experience no noticeable delays.

[kizer]

I personally don't have any delays or errors. I'm using the following:

Linux XMBC Live Installed 9+

All Nfo's stored on server along with thumbnails, which is then cached to machine

cache_dirs

I'm not using any bizarre SQL mods or configurations.

 

When my drives are not spun up my machine exhibits a bit of a glitch while the buffering happens, but that normally only takes a second or two if I start something but it does play and it shows "Buffering" across the top.

 

In My Opinion cache_dirs is a must for XBMC. I ran it before without it and honestly it wasn't as responsive and waiting for drives to spin up to scan my server was a bit annoying. LOL

[PeterB]

When my drives are not spun up my machine exhibits a bit of a glitch while the buffering happens, but that normally only takes a second or two if I start something but it does play and it shows "Buffering" across the top.

 

Mine does not do this.  Are you using NFS or SMB?

[kizer]

I'm using SMB, but what I should of said

Day to day use I don't have any issues what so ever. If I start a show and then stop it, turn off XBMC and my unRAID server turns off. If I come back to it say a few days later and then resume my show its buffered on my unRAIDs drive cache then it gives me that buffering if a drive is off. I'm guessing thats normal since you can't expect a drive to magically give you data if it was off in the first place.

 

Normal operation it runs smooth as can be.