Kodi Community Forum

Morning All


A rather strange one here. I have a back end server running Ubuntu Karmic with two usb (yes, I know) external 1TB drives with media.These are set up as shares in samba. My samba install has aio set and is pretty rock solid for access.

My front end is a revo running karmic with xbmc (last final) and I've solved most of the issues with the box bar one.

The revo is hard ethernet connected to a switch that also connects to the server everything is fine for 100mb.

I've tested the file transfer speed in system file tool within xbmc for files and everything seems fine.

Now for the problem.

On playback of media I get glitches. These do not increase the dropped frame count but when I check in the log I find these

Quote:19:11:26 T:2929863536 M:1151926272 ERROR: Read - Error( -1, 22, Invalid argument ) - Retrying

Now the REALLY weird thing is these start happening every 5 minutes, so I'll see one at 19:11, one at 19:16, one at 19:21 and so on.

I've checked my Samba log.nmbd file, log.kraken (the revo box) syslog and messages to see if anything coincides with these issues on the server side and I can't find anything.

I've done a similar exercise

Quote:cd /var/log
sudo fgrep -yR 19:11 * | grep -y "sep 30"

On the revo and the only thing there is the wifi linux wpa_supplicant process reporting back but that happens every minute or so due to a bug in the process anyway.

Sooooo I'm a little stuck. Anyone got any ideas, happy to post a pastebin of any logs if it will be diagnosticly helpful. or am I up the wrong tree completely and a read error isnt file read but something else.

Also has anyone seen this issue before and is it fixed by dharma?

Many thanks all


Edz

...aaaand updating and having a quick bump here i'll add that i'm going to try adding

Quote:dead time = 0

to the global section of my smb.conf in case connections are dropping mid stream and the hiccough is the connection being remade.

Bump again

Ed

Nope

not sure what this is but I dont think its samba.

Any help, please?

Quick bump, no one have any ideas?

Devs, what does the "raw" error message mean when dumped into the log?

Thanks

Hmm copying the file to the revo plays ok. This is definitely network related either at the revo end or Samba. I think deadtime should be one word though will try that this evening.

No one else had this issue then?
Anyone, Bueller?

An update on my continuing odyssey.

In order to rule out (or in) problems with the samba share I’ve amended my back end server to share via NFS, adding the media share folders into the /etc/exports. I’ve then added the NFS client and added the new shares into /etc/fstab on the Revo.

Then came the painful bit.

I thought it would be simple, opened up each source in videos and amended it to point from

Quote:smb://192.168.0.XXX/media/NAS/shared/Video/ …. Etc
to be
/home/media/NAS/shared/Video/ … Etc

However, in XBMC (Camelot at least) this does not amend existing entries. So I had to clear the library and then re-import (I export to separate files / actors etc so this wasn’t too painful just time consuming)

By this time it was weeeeell after 2AM so I only had a quick check but xbmc.log did confirm it is using /home/media as the source (the mount point for the NFS share)

This evening I will try some watching (preferably with a glass of something) and then see (A) if I get jumps and (B) any read errors occur in the log.

If nothing else I’ve got NFS working which was less painful than I imagined.

So if I get read errors I know it wasn’t samba and its either the networking chain or the server file system / USB drives (USB 2.00) to blame (I didn’t see this on local playback when tested with the Revo) ..

The story continuums….

I personaly use only nfs if possible.
Even with Windows 7 you have a nfs-client included and nfs was allways faster than a smb share.

linuxluemmel Wrote:I personaly use only nfs if possible.
Even with Windows 7 you have a nfs-client included and nfs was allways faster than a smb share.

I always thought it would pose more issues, it certainly did the last time I tried it but that box was an ages old machine with an early build of knoppmyth debian on it, the newer ubuntu repositries make life a lot easier!

I have a mix of osx and windows 7 and xp as well as linux boxes and android mobiles using my home network so keeping the samba shares and adding NFS seemed a logical choice.

We'll see what, if anything, it does to stop the errors this evening!

If you dont mind me asking, what setup are you using?

And updating some more..

First off, playing back some tv episodes and even the first 20 minutes of a “troublesome” converted region 1 dvd which almost always bounced about…

… we had no read error messages. I though I saw a small jump but this could just now be a psychosis since there was nothing in xbmc.log about read errors.

An interesting issue I did run into involves the back end configuration of NFS on the server though. The media drives are Seagate USB 1TB units, these are external since they can then be unplugged and stored securely if we go away without affecting the servers mythTV duties (myth uses an internal storage drive as does the OS).

Since Seagates have the notorious spindown under linux issue I had originally used sdparm to turn this off under my old knoppmyth installation, when I moved to ubuntu this was no longer an issue so I reinstated the spindown. Under NFS with these drives in the revos fstab it can cause the Ethernet connection to fail or at least be delayed at boot so I’ve had to clear spindown again for now, I’ll continue taking a look at this.

The other issue was that all drives would mount on the Revo but whichever drive wasn’t being used (say whilst watching a movie from the second NAS drive) would fall victim to inactivity and I’d get an NFS stale file handle issue (the files would no longer be available).

I’ve fixed that one by changing the export file to include fsid options as I found here

http://blog.wpkg.org/2007/10/26/stale-nfs-file-handle/

The saga continues….

… although I may have a night off tonight