rhythmbox keeps dropping smb connection to nas device

[Andy Stevens]

I have a fresh installation of the 12.04 beta running from USB drive (created with usb-creator and the ISO) on an Acer Aspire One netbook, connecting to my home network via wifi.
I have a dozen or so albums on a network drive that I previously ripped from CD to FLACs. I can connect to the smb share okay in nautilus and browse them, and I have set rhythmbox to use this location (smb://...) for the music library. I have the "scan for new files" preference enabled.
When I start up rhythmbox it begins scanning the files on the remote drive and adding them to the library. All is well for a short while, then all of a sudden it starts showing an Import Error for every file, and I eventually end up with only a fraction of the music files showing in the library.

It appears that for some reason the SMB connection is dropped; when I switched back to Nautilus, the tab that was showing one of the album folders has been closed and the network connection for that share no longer shows in the sidebar. I didn't see the wifi connection drop & reconnect, and the connection to another share on the same NAS drive was still showing in Nautilus, so it looks like it was only this particular share that got unmounted rather than something more general.
Unfortunately, after reconnecting to the share in nautilus, there's no obvious way to retry importing the missing files (e.g. a Retry option when right-clicking on one of the errors or Missing Files?) so I quit and restart rhythmbox. At which point, exactly the same happens again... If I don't run rhythmbox, it stays connected to the share and I can browse as long as I like in nautilus without any problem. But every time I start rhythmbox the connection dies shortly afterward, during the library scan.
I'm not sure whether the actual problem is with rhythmbox itself, or just that the level of load produced by the library scan triggers some problem in the SMB client code.

I had a similar but less severe problem when I originally ripped the CDs using 12.04 Milestone 1; it usually took a few goes to get right through each CD, as at some point it would stop with an error having lost the connection to the share and left a zero-byte file in the folder. After reconnecting and viewing the folder using Nautilus (to see how far it had got) I set it going again on the remaining tracks (overwriting the truncated file) and eventually completed the album. I hadn't observed the "library scan" problem with the earlier milestone, then again I hadn't tried starting over with an empty library in the earlier release either.
---
ApportVersion: 1.94.1-0ubuntu2
Architecture: i386
CasperVersion: 1.311
DistroRelease: Ubuntu 12.04
LiveMediaBuild: Ubuntu 12.04 LTS "Precise Pangolin" - Alpha i386 (20120317)
Package: rhythmbox
PackageArchitecture: i386
ProcEnviron:
 TERM=xterm
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 3.2.0-18.29-generic-pae 3.2.9
Tags: precise
Uname: Linux 3.2.0-18-generic-pae i686
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups:

[Vadim Rutkovsky (roignac)]

Thank you for taking the time to report this bug and helping to make
Ubuntu better, however the report is missing some information, may you
please run rhythmbox as: rhythmbox --debug &> rhythmbox-debug.txt ;
perform the steps to reproduce the bug and attach that resulting log
file to the report? Thanks in advance!

[Andy Stevens (stevensa)]

Here's the log you asked for. Part way through it starts showing errors "The specified location is not mounted".
Also, the authentication dialog appeared repeatedly wanting me to log in to the share again and again, despite ticking the box to remember the details until logout.

[Vadim Rutkovsky (roignac)]

Looks like gvfsd has died.

[Andy Stevens (stevensa)]

Tried it again just now; when it lost the connection dmesg revealed
[ 3539.119471] gvfsd-smb[15727]: segfault at 14 ip b7177373 sp bfc7cb50 error 4 in libgvfsdaemon.so[b7169000+23000]
at the end of the log. Looks like you're right that it's a gvfsd problem. I've only observed it with rhythmbox; on the other hand I've probably not had anything else making heavy use of a share while running 12.04, just the occasional file copy of a podcast.

[Andy Stevens (stevensa)]

Just tried again with the latest daily CD image, but I can see a similar message in dmesg
[12240.022073] gvfsd-smb[7228]: segfault at 14 ip b7149373 sp bfffd610 error 4 in libgvfsdaemon.so[b713b000+23000]
I'll attach the ubuntu-bug output for the gvfs-libs package, which contains that .so file.
(I had to use a daily build, as before ubuntu-bug was complaining about out of date packages, but this memory stick didn't have enough room on it to apply the current updates...)

[Andy Stevens (stevensa)]

apport information

[Andy Stevens (stevensa)]

I also tried with an 11.04 usb stick that I still had lying around. Works perfectly in that, 140-odd tracks all imported with no errors. Can't try 11.10 as the wifi doesn't work on my netbook (Acer Aspire One ZG5/110-Ab) in that release.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Jens Finkhäuser (finkhaeuser-consulting)]

Sound plays back for a split second every time I unpause it, and then goes mute. Video playback from the same share via totem works fine, so probably not a networking-related problem.

The same machine played music just fine from the same share before upgrading from oneiric to precise.

[Jens Finkhäuser (finkhaeuser-consulting)]

Right-click and "open with GNOME Mplayer" plays the same tracks just fine FWIW.

[Walter Garcia-Fontes (walter-garcia)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. We are sorry that we do not always have the capacity to look at all reported bugs in a timely manner. There have been many changes in Ubuntu since that time you reported the bug and your problem may have been fixed with some of the updates. It would help us a lot if you could test it on a currently supported Ubuntu version. When you test it and it is still an issue, kindly upload the updated logs by running apport-collect <bug #> and any other logs that are relevant for this particular issue.