[block134]

I take back my prior statement, if I look at the correct log I do see the errors in it. Here is part of my log http://pastebin.com/Hn28FNnz, pastebin wouldn't let me put it all in there is if you want the rest just let me know.

[Memphiz]

issue should be fixed in the next nightly ...

[User 34959]

So this might be an issue when scraping an nfs source while pathsubstitution via nfs is used - anyone thinks else?

That seems to be the case. I'm using Dec 10th build on OSX, video sources set using nfs and mysql db. Using nfs for path substitution results in extremely poor performance navigation wise and thumbnail loading wise (takes 10 seconds to enter a movie library with less than 100 titles). Scraping seems to get randomly interrupted. As soon as i used smb for path substation everything flies..

I'll try the Dec 11th build next.

[vajonam]

That seems to be the case. I'm using Dec 10th build on OSX, video sources set using nfs and mysql db. Using nfs for path substitution results in extremely poor performance navigation wise and thumbnail loading wise (takes 10 seconds to enter a movie library with less than 100 titles). Scraping seems to get randomly interrupted. As soon as i used smb for path substation everything flies..

I'll try the Dec 11th build next.

http://forum.xbmc.org/showthread.php?tid=116805

I saw this with build on the 12th, dont know if its realted.

[renhack]

Bringing this back for clarification.

I'm experiencing the same issues ever since I switched to Eden Beta. I use NFS for my sources with MySQL. I also use NFS for my thumbnail path substitution. Every time i try to scrape or just browse my file list i lose connection to my NFS shares within 10-30 seconds.

Is the solution to this just switching the path substitution back to SMB for the thumbs? I really don't want to swap it out just to try since I've already gone through the hassle of dropping NFS and rescanning my library. I'd love NFS back though, it was definitely faster. Thanks.

[Memphiz]

Bringing this back for clarification.

I'm experiencing the same issues ever since I switched to Eden Beta. I use NFS for my sources with MySQL. I also use NFS for my thumbnail path substitution. Every time i try to scrape or just browse my file list i lose connection to my NFS shares within 10-30 seconds.

Is the solution to this just switching the path substitution back to SMB for the thumbs? I really don't want to swap it out just to try since I've already gone through the hassle of dropping NFS and rescanning my library. I'd love NFS back though, it was definitely faster. Thanks.

New xbmc.log needed with debugging turned on please. Also what NFS server are you using?

[renhack]

I posted all of this in another thread here. http://forum.xbmc.org/showthread.php?tid=117976 It should all be relevant still.

I have 2 other ATV2's running on pre-eden from 10/16 Nightly with no NFS issues. I did a factory restore and fresh install of Eden Beta. NFS works for about 30 seconds, then it disconnects. The strange thing is that it starts working again after NFS goes idle and closes the connections. You can see all of this in the log i post at the end.

My setup is 3 ATV's running on MYSQL. I'm using WHS2011 as my server with built in NFS. My NFS share setting are..

- Everyone group has full control
- Kerberos V5 integrity and authentication Krb5i checked
- Kerberos V5 authentication Krb5 checked
- No server authentication checked with unmapped access by UID
- NFS permission are read/write with root access ANSI encoding

NFS Settings


NFS Permissions


advancedsettings.xml - http://pastebin.com/mP6ivCBc
sources.xml - http://pastebin.com/vYRer8FT
mediasources.xml - http://pastebin.com/3P73Gz6f

XBMC Log - http://pastebin.com/S03PyYF6
Syslog - http://pastebin.com/dLue0B1c

Hopefully this is enough information to debug this. Thanks

[Memphiz]

Could you try to merge media-nfs1 and xbmc-nfs1 by copying the thumbnail folder into xbmc-nfs1 and change the pathsubstitution so that only one export is used all over. This would prevent the need for remounting the exports a dozens of times. (I have to refactor NFS and bump version when i get some more time where i want to rework the handling for preventing this too).

Maybe its just that the server gets to much mount requests and limits them somehow.

[renhack]

My initial thought was that it was being limited but I can't find any error reporting on the server side and the XBMC log just seems to show the point where it stops working. There must have been a change in how XBMC handles NFS since I don't have this issue on my October nightly builds. I wish I would have kept up with nightlys better so I would have known when the issue started.

I switched to smb for now, but its not ideal since NFS seemed to solve some buffering issues in the past. I'll try your single share solution with my dev atv over the weekend.

[Memphiz]

yeah test it ... if that works for you ... then this commit will take care of it in beta 2:

https://github.com/xbmc/xbmc/commit/f57c...8c66871d5a

if it doesn't this commit is just another improvement for nfs

[renhack]

I finally found some solid info on Windows "Services for NFS". Looks like the default limitations might be the culprit.

16 TCP connections per IP address
32 Default Concurrent Connections

If anyone else runs into the same issue, try editing the registry. These are max values. You may want to play with something in between.

Code:

[b]Default Concurrent Connections[/b]
HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Rpcxdr\Parameters

Right-click DefaultConcurrentConnections, and then click Modify to 1024

Code:

[b]TCP connections per IP address[/b]
HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Rpcxdr\Parameters

Right-click MaxConcurrentConnectionsPerIp, and then click Modify to 256

[Memphiz]

k ... then i guess this issue will be fixed in next beta then ...