Bug#714736: Extra waiting caused by NFS4_DELAY
First of all, a great thank you Jan-Marek for pointing out the (very
probable) root cause of the problem.
First, I tried to mount using minorversion=1. Unfortunately, NFS4.1
appears not to be supported:
root@bural:/home/sylvain# mount -t nfs4 real.chicoree.fr:/homes /mnt -o
sec=krb5,minorversion=1
mount.nfs4: Protocol not supported
Both client server having the same packages (Wheezy + NFS/linux-image
from Jessie):
ii libnfsidmap2:amd64 0.25-4 amd64 NFS idmapping library
ii linux-image-3.9-1- 3.9.6-1amd64 Linux 3.9 for
64-bit PCs
ii nfs-common 1:1.2.8-4 amd64 NFS support files
common to client and se
ii nfs-kernel-server 1:1.2.8-4 amd64 support for NFS
kernel server
What is the minimum kernel/package version required to use NFS4.1 on Debian?
---
After that, I tried to stick to NFS4.0, having both gssd and svcgssd
running on the client and the server. So I installed
nfs-{common,kernel-server} on both machines.
But, just installing nfs-kernel-server and editing
/etc/defaults/nfs-kernel-server to force execution of rpc.svcgssd is not
sufficient on the client, since the init script won't run rpc.svcgssd if
there is no export.
So, I manually ran svcgssd on the client and gssd on the server (real
is the server. bural the client):
root@real:~# ps -edf | grep gssd
root 3753 1 0 Jul08 ?00:00:00 /usr/sbin/rpc.svcgssd -vvv
root 5254 5193 0 14:58 pts/000:00:00 grep gssd
root@real:~# /usr/sbin/rpc.gssd -fvvv
beginning poll
destroying client /var/lib/nfs/rpc_pipefs/nfsd4_cb/clntf
root@bural:/home/sylvain# ps -edf | grep gss
root 14146 1 0 14:46 ?00:00:00 /usr/sbin/rpc.gssd -vvv
root 14246 14142 0 15:00 pts/500:00:00 grep gss
root@bural:/home/sylvain# /usr/sbin/rpc.svcgssd -fvvv
entering poll
As far as I can tell, both GSS servers are running on both machines.
However, this does not appear to solve the issue. Despite -vvv option,
I see no informations. But I still have the usual messages on server's
syslog:
Jul 10 15:29:00 real kernel: [166575.902975] RPC: AUTH_GSS upcall timed out.
Jul 10 15:29:00 real kernel: [166575.902975] Please check user daemon is
running.
Probably, there is there something I missed while launching the gss
services? Maybe there is some other services to restart?
And now for something different:
The attached patch improves Nicolas version for the cases of NFS version
!= 4.0, where svcgssd is not needed.
This is probably a corner case, but I'm not sure Nicolas' patch will
behave as expected when the client import both 4.0 and 4.x shares:
nfs40.server:/this /this nfs4 rw,sec=krb5 0 0
nfs41.server:/that /that nfs4 rw,sec=krb5,minorversion=1 0 0
-- Sylvain
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#714736: ... use NFS 4.1
So you can use the mount option minorversion=1, like
mount -t nfs4 -o minorversion=1 ...
Continuing my investigations on that topic, it appears that nfs-utils is
build with NFS 4.1 support since at least 1.2.6-3:
https://buildd.debian.org/status/fetch.php?pkg=nfs-utilsarch=amd64ver=1%3A1.2.6-3stamp=1341956993
dpkg-buildpackage: source package nfs-utils
[...]
CFLAGS=-g -Wall -O2 ./configure \
--mandir='${prefix}/share/man' \
--enable-nfsv41 \
--enable-ipv6 \
--enable-libmount-mount \
--with-tcp-wrappers
... so are the Debian kernels for Jessie and Wheezy:
sylvain@bural:~$ cat /boot/config-3.2.0-4-amd64 | grep _NFS_V
CONFIG_NFS_V3=y
CONFIG_NFS_V3_ACL=y
CONFIG_NFS_V4=y
CONFIG_NFS_V4_1=y
sylvain@bural:~$ cat /boot/config-3.9-1-amd64 | grep _NFS_V
CONFIG_NFS_V2=m
CONFIG_NFS_V3=m
CONFIG_NFS_V3_ACL=y
CONFIG_NFS_V4=m
CONFIG_NFS_V4_1=y
CONFIG_NFS_V4_1_IMPLEMENTATION_ID_DOMAIN=kernel.org
But strangely /proc/fs/nfsd/versions seems to disagree with that (both
on client and server):
root@real:~# cat /proc/fs/nfsd/versions
+2 +3 +4 -4.1
More (?) strangely, /var/log/syslog shows some sign of NFS 4.1 support:
Jul 10 16:05:54 real kernel: [ 11.126765] RPC: Registered tcp NFSv4.1
backchannel transport module.
At this point, I'm quite puzzled. I have the strange feeling that
NFSv4.1 + Krb5 should work -- but somehow don't on our network ?!?
Either there is really a bug somewhere - or there is some kind of oddity
on our machines' configuration.
Could someone confirm successful deployment of NFSv4.1 + Krb5 on Debian
Wheezy -- so I would stop chasing ghost bugs (and stop bothering you at
the same time).
-- Sylvain
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#648155: nfs-common: nfs mount hangs when kerberos ticket expires.
The bug is still present on latest stable wheezy. This is quite annoying since this make NFS4+Kerberos almost unusable where users have the opportunity to put their computer to sleep -- and the Kerberos ticket expires during the sleep period. If I understand it well, this is a kernel bug. Is there any hope the proposed patch made its way in Wheezy's kernel? As a side node, is this a duplicate of http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=446238 ? -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#714736: [Pkg-xfce-devel] Bug#714736: Bug#714736: thunar: GUI freeze while moving to trash a file from a NFS4+Kerberos mount
On 04/07/2013 22:52, Yves-Alexis Perez wrote: It'd help if you could try to reproduce with no thunar at all. Just try to move the same set of files to the trash folder. Using mv to move files to .local/share/Trash/files does *not* exhibit that behavior: # This works like a charm: sh$ touch 1 2 3 4 sh$ mv 1 .local/share/Trash/files sh$ mv 2 .local/share/Trash/files sh$ mv 3 .local/share/Trash/files ... As far as I can tell, move to trash does some more work than only moving the file. At bare minimum it records the original path an deletion date in the corresponding .trashinfo files located in .local/share/Trash/info/ . Nevertheless ... # ... using the trash-put command from trash-cli package # works too: sh$ touch 1 2 3 4 sh$ trash-put 1 sh$ trash-put 2 sh$ trash-put 3 ... # But this hang for each file... sh$ touch 1 2 3 4 -- from Thunar, drag'n drop 1 to Trash -- from Thunar, click on 2, press delete -- -- Sylvain Leroux -- sylv...@chicoree.fr -- http://www.chicoree.fr -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#714736: [Pkg-xfce-devel] Bug#714736: thunar: GUI freeze while moving to trash a file from a NFS4+Kerberos mount
I've done some more tests (on pure wheezy machine). The bug is not as systematic as I thought at first. Sometimes the user is able to move several files to trash without any difficulties. But for no apparent reason, for some file the process take long long time. The problem appears both while configuring the NFS share as sync or async. But, remounting /home as NFS without Kerberos appears to solve the problem. By looking in /var/log/syslog on the NFS/Kerberos server it appears that, when user experiences a hang, I have the following messages (up to 4 times): Jul 4 10:56:40 real kernel: [68746.902591] RPC: AUTH_GSS upcall timed out. Jul 4 10:56:40 real kernel: [68746.902593] Please check user daemon is running. Jul 4 10:56:58 real kernel: [68764.934597] RPC: AUTH_GSS upcall timed out. Jul 4 10:56:58 real kernel: [68764.934600] Please check user daemon is running. The timestamps are coherent with the user experienced delay. I checked, rpc.svcgssd is running on the server. I tried to kill it and the manually run it with -vvv -iii -rrr to gather more informations -- but nothing at all is logged when the problem appears. I'm running nfs-common 1:1.2.6-4 on the server I think this bug is not related to thunar, but to nfs-common -- and it should be reassigned (I don't know how to do that). Regards - Sylvain -- -- Sylvain Leroux -- sylv...@chicoree.fr -- http://www.chicoree.fr -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#714863: rhythmbox: Missing dependency on gnome-icon-theme-symbolic
Package: rhythmbox Version: 2.97-2.1 Severity: minor Dear Maintainer, rhythmbox requires the package gnome-icon-theme-symbolic to proper display search (magnifying glass) and play/pause mini-icons in the main GUI window - Sylvain -- System Information: Debian Release: 7.1 APT prefers stable-updates APT policy: (500, 'stable-updates'), (500, 'stable') Architecture: amd64 (x86_64) Kernel: Linux 3.2.0-4-amd64 (SMP w/4 CPU cores) Locale: LANG=fr_FR.UTF-8, LC_CTYPE=fr_FR.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Versions of packages rhythmbox depends on: ii dbus 1.6.8-1+deb7u1 ii gnome-icon-theme 3.4.0-2 ii gstreamer0.10-plugins-base 0.10.36-1.1 ii gstreamer0.10-plugins-good 0.10.31-3+nmu1 ii gstreamer0.10-x 0.10.36-1.1 ii libatk1.0-0 2.4.0-2 ii libc62.13-38 ii libcairo-gobject21.12.2-3 ii libcairo21.12.2-3 ii libffi5 3.0.10-3 ii libgdk-pixbuf2.0-0 2.26.1-1 ii libgirepository-1.0-11.32.1-1 ii libglib2.0-0 2.33.12+really2.32.4-5 ii libgstreamer-plugins-base0.10-0 0.10.36-1.1 ii libgstreamer0.10-0 0.10.36-1.2 ii libgtk-3-0 3.4.2-6 ii libice6 2:1.0.8-2 ii libjson-glib-1.0-0 0.14.2-1 ii libpango1.0-01.30.0-1 ii libpeas-1.0-01.4.0-2 ii librhythmbox-core6 2.97-2.1 ii libsm6 2:1.2.1-2 ii libsoup-gnome2.4-1 2.38.1-2 ii libsoup2.4-1 2.38.1-2 ii libtdb1 1.2.10-2 ii libtotem-plparser17 3.4.2-1 ii libxml2 2.8.0+dfsg1-7+nmu1 ii media-player-info17-1 ii rhythmbox-data 2.97-2.1 ii zlib1g 1:1.2.7.dfsg-13 Versions of packages rhythmbox recommends: ii avahi-daemon 0.6.31-2 ii gstreamer0.10-plugins-ugly 0.10.19-2+b2 pn gstreamer0.10-pulseaudio none ii gvfs-backends1.12.3-4 ii notification-daemon 0.7.6-1 ii rhythmbox-plugins2.97-2.1 ii xfce4-notifyd [notification-daemon] 0.2.2-2 ii yelp 3.4.2-1+b1 Versions of packages rhythmbox suggests: pn gnome-codec-install none pn gnome-control-center none ii gstreamer0.10-plugins-bad0.10.23-7.1 pn rhythmbox-plugin-cdrecorder none -- no debconf information -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#714736: [Pkg-xfce-devel] Bug#714736: thunar: GUI freeze while moving to trash a file from a NFS4+Kerberos mount
On 02/07/2013 21:42, Yves-Alexis Perez wrote: On mar., 2013-07-02 at 13:07 +0200, Sylvain Leroux wrote: I think the exact same bug has been reported some time ago for Ubuntu: https://bugs.launchpad.net/ubuntu/+source/thunar/+bug/1051952 According to that, the bug could be traced back to `libgio` call `g_file_trash()` Could you try in more recent Thunar and see if it still applies there? Regards, Exact same behavior upgrading thunar (incl. dependencies) to the version available on Jessie: ii libglib2.0-0:amd642.36.1-2build1 ii libglib2.0-data 2.33.12+really2.32.4-5 ii libglibmm-2.4-1c2a:amd64 2.32.1-1 ii libthunarx-2-01.6.3-1 ii thunar1.6.3-1 ii thunar-data 1.6.3-1 -- -- Sylvain Leroux -- sylv...@chicoree.fr -- http://www.chicoree.fr -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#714736: [Pkg-xfce-devel] Bug#714736: thunar: GUI freeze while moving to trash a file from a NFS4+Kerberos mount
On 02/07/2013 21:42, Yves-Alexis Perez wrote: On mar., 2013-07-02 at 13:07 +0200, Sylvain Leroux wrote: I think the exact same bug has been reported some time ago for Ubuntu: https://bugs.launchpad.net/ubuntu/+source/thunar/+bug/1051952 According to that, the bug could be traced back to `libgio` call `g_file_trash()` Could you try in more recent Thunar and see if it still applies there? Regards, ... and upgrading gvfs (I think (?) this manage the trash somehow) has no effect neither. ii gvfs:amd64 1.16.3-1 amd64 -- -- Sylvain Leroux -- sylv...@chicoree.fr -- http://www.chicoree.fr -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#714736: thunar: GUI freeze while moving to trash a file from a NFS4+Kerberos mount
Package: thunar Version: 1.2.3-4+b1 Severity: important Dear Maintainer, Users experience a GUI freeze when they move to trash a file with Thunar on a machine where /home is remotely mounted using NFS4 + Kerberos authentication. For something like 30 seconds the GUI is not responsive (can't launch other apps, can move windows, but there is no window redraw). After that finally the move-to-trash succeed. It seems to only appears on user /home directory. This is the only NFS mount on the system. Here is the corresponding /etc/fstab entry: ---8 # /etc/fstab real.chicoree.fr:/homes /home nfs4 rw,sec=krb5 0 0 ---8 Here are some of the corresponding packages that might be involved: ii krb5-config 2.3 ii krb5-locales1.10.1+dfsg- ii krb5-user 1.10.1+dfsg- ii libgssapi-krb5 1.10.1+dfsg- ii libkrb5-3:amd6 1.10.1+dfsg- ii libkrb5support 1.10.1+dfsg- ii libnfsidmap2:a 0.25-4 ii libpam-krb5:am 4.6-1 ii libthunarx-2-0 1.2.3-4+b1 ii libxfce4ui-1-0 4.8.1-1 ii libxfce4util-b 4.8.2-1 ii libxfce4util-c 4.8.2-1 ii libxfce4util4 4.8.2-1 ii libxfcegui4-4 4.8.1-5 ii nfs-common 1:1.2.6-4 ii thunar 1.2.3-4+b1 ii thunar-data 1.2.3-4 ii xfce-keyboard- 4.8.1-1 ii xfce4 4.8.0.3 ii xfce4-appfinde 4.8.0-3 ii xfce4-notifyd 0.2.2-2 ii xfce4-panel 4.8.6-4 ii xfce4-session 4.8.3-3 ii xfce4-settings 4.8.3-2 ii xfce4-terminal 0.4.8-1+b1 ii xfce4-utils 4.8.3-2 ii xfce4-volumed 0.1.13-3 ii xfdesktop4 4.8.3-2 ii xfdesktop4-dat 4.8.3-2 Client machines are running stable Debian Wheezy. For now, there is no workaround (beyond just wait -- or not using NFS4+Kerberos, which is unacceptable in our case). This has an important impact on usability of the system for the end user since the bug is regularly in the way during day to day activities. -- System Information: Debian Release: 7.1 APT prefers stable-updates APT policy: (500, 'stable-updates'), (500, 'stable') Architecture: amd64 (x86_64) Kernel: Linux 3.2.0-4-amd64 (SMP w/4 CPU cores) Locale: LANG=fr_FR.UTF-8, LC_CTYPE=fr_FR.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Versions of packages thunar depends on: ii desktop-file-utils 0.20-0.1 ii exo-utils 0.6.2-5 ii libatk1.0-0 2.4.0-2 ii libc6 2.13-38 ii libcairo2 1.12.2-3 ii libdbus-1-3 1.6.8-1+deb7u1 ii libdbus-glib-1-20.100.2-1 ii libexo-1-0 0.6.2-5 ii libgdk-pixbuf2.0-0 2.26.1-1 ii libglib2.0-02.33.12+really2.32.4-5 ii libgtk2.0-0 2.24.10-2 ii libgudev-1.0-0 175-7.2 ii libice6 2:1.0.8-2 ii libnotify4 0.7.5-1 ii libpango1.0-0 1.30.0-1 ii libsm6 2:1.2.1-2 ii libthunarx-2-0 1.2.3-4+b1 ii libxfce4ui-1-0 4.8.1-1 ii libxfce4util4 4.8.2-1 ii shared-mime-info1.0-1+b1 ii thunar-data 1.2.3-4 Versions of packages thunar recommends: ii dbus-x111.6.8-1+deb7u1 ii gvfs1.12.3-4 ii libfontconfig1 2.9.0-7.1 ii libfreetype62.4.9-1.1 ii thunar-volman 0.6.1-1 ii tumbler 0.1.25-1+b1 ii xdg-user-dirs 0.14-1 ii xfce4-panel 4.8.6-4 Versions of packages thunar suggests: pn thunar-archive-plugin none pn thunar-media-tags-plugin none -- no debconf information -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#714736: thunar: GUI freeze while moving to trash a file from a NFS4+Kerberos mount
I think the exact same bug has been reported some time ago for Ubuntu: https://bugs.launchpad.net/ubuntu/+source/thunar/+bug/1051952 According to that, the bug could be traced back to `libgio` call `g_file_trash()` -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#682709: NFS4 krb5 mounts hang under nfs-utils 1.2.6-3
I confirm the exact same behavior with a fresh install of wheezy (stable) with nfs4+MIT kerberos5. Clients can no longer mount nfs shares and hangs forever. By looking at the network traffic with Wireshark, it appears that the server never send response to the client query NFS V4 COMPOUND Call PUTROOTFH;GETFH;GETATTR Like Julien Blanc (post #51), the client name / ip are in the /etc/hosts of the NFS+Krb5 server. But, in my case, installing the binary packages provided by Daniel Hess (post #44) solved the problem. Best regards, - Sylvain Leroux -- -- Sylvain Leroux -- sylv...@chicoree.fr -- http://www.chicoree.fr -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
