Red Hat Bugzilla – Bug 470706
beagled uses up all inotify watches
Last modified: 2009-12-18 01:48:02 EST
Created attachment 322988 [details] typescript of gdb session, attempting to run gpk-update-viewer +++ This bug was initially created as a clone of Bug #456513 +++ Description of problem: After running a Gnome app based on PolicyKit a number of times, the app stops working. Some affected apps: - gpk-application - gpk-update-viewer - polkit-gnome-example Version-Release number of selected component (if applicable): PolicyKit-0.8-2.fc9.x86_64 PolicyKit-0.8-2.fc9.i386 PolicyKit-gnome-0.8-4.fc9.x86_64 How reproducible: Run polkit-based app a number of times. Steps to Reproduce: 1. 2. 3. Actual results: App crashes with following backtrace: [WARN 22576] polkit-error.c:143:polkit_error_get_error_message(): error != NULL Not built with -rdynamic so unable to print a backtrace ** (polkit-gnome-example:22576): WARNING **: Failed to initialize PolicyKit context: (null) [WARN 22576] polkit-error.c:156:polkit_error_free(): error != NULL Not built with -rdynamic so unable to print a backtrace [WARN 22576] polkit-error.c:143:polkit_error_get_error_message(): error != NULL Not built with -rdynamic so unable to print a backtrace ** (polkit-gnome-example:22576): WARNING **: Failed to initialize PolicyKit context: (null) [WARN 22576] polkit-error.c:156:polkit_error_free(): error != NULL Not built with -rdynamic so unable to print a backtrace (polkit-gnome-example:22576): GLib-GObject-WARNING **: invalid (NULL) pointer instance (polkit-gnome-example:22576): GLib-GObject-CRITICAL **: g_signal_connect_data: assertion `G_TYPE_CHECK_INSTANCE (instance)' failed (polkit-gnome-example:22576): GLib-GObject-WARNING **: invalid (NULL) pointer instance (polkit-gnome-example:22576): GLib-GObject-CRITICAL **: g_signal_connect_data: assertion `G_TYPE_CHECK_INSTANCE (instance)' failed Expected results: App runs the same as normally. Additional info: --- Additional comment from dennis.wagelaar@vub.ac.be on 2008-07-24 06:25:50 EDT --- Workaround: Log out and back in: PolicyKit-based apps are working again --- Additional comment from aravind.parchuri@gmail.com on 2008-08-27 21:06:03 EDT --- I can confirm this bug - apps like gnome-system-monitor, the calendar applet, etc start crashing after a while. Here's the backtrace in case it helps: ** (gnome-system-monitor:13931): WARNING **: Failed to initialize PolicyKit context: (null) [WARN 13931] polkit-error.c:156:polkit_error_free(): error != NULL Not built with -rdynamic so unable to print a backtrace ---Type <return> to continue, or q <return> to quit--- Program received signal SIGSEGV, Segmentation fault. 0x02fbc5f2 in ?? () from /usr/lib/libpolkit-gnome.so.0 (gdb) bt #0 0x02fbc5f2 in ?? () from /usr/lib/libpolkit-gnome.so.0 #1 0x02fbe938 in ?? () from /usr/lib/libpolkit-gnome.so.0 #2 0x00369991 in ?? () from /lib/libgobject-2.0.so.0 #3 0x02fbe99b in ?? () from /usr/lib/libpolkit-gnome.so.0 #4 0x0036a0de in g_object_newv () from /lib/libgobject-2.0.so.0 #5 0x0036acd2 in g_object_new_valist () from /lib/libgobject-2.0.so.0 #6 0x0036ae1e in g_object_new () from /lib/libgobject-2.0.so.0 #7 0x02fbcfe8 in polkit_gnome_action_new_default () from /usr/lib/libpolkit-gnome.so.0 #8 0x08055577 in gtk_toggle_button_set_active () #9 0x08054528 in gtk_toggle_button_set_active () #10 0x009b15d6 in __libc_start_main () from /lib/libc.so.6 #11 0x08052bb1 in gtk_toggle_button_set_active () (gdb) --- Additional comment from aravind.parchuri@gmail.com on 2008-08-27 21:07:41 EDT --- BTW, I'm on i386, not x86_64. --- Additional comment from rhughes@redhat.com on 2008-08-28 04:17:21 EDT --- Have you got a 100% way of reproducing this? I can't see to reproduce on i386 or x64. --- Additional comment from aravind.parchuri@gmail.com on 2008-08-28 15:18:53 EDT --- I don't really have a step-by-step procedure to reproduce this. I'm pretty sure that a few hours after logging into a gnome session, some apps start crashing with this error, consistently. I'd be happy to help debug. Is there some policykit backend that I should monitor? Any locks that might be stuck? Maybe I'll just snapshot /var/lib/PolicyKit* before and after the problem and attach them if they don't contain any personal info. --- Additional comment from dennis.wagelaar@vub.ac.be on 2008-09-01 10:21:27 EDT --- This does happen 100% of the time for me on both of my Fedora machines at home. Both of those machines were upgraded from Fedora 8 using yum (http://fedoraproject.org/wiki/YumUpgradeFaq). --- Additional comment from john@redux.org.uk on 2008-10-15 12:47:26 EDT --- This is happening on my f9 x86_64 box that I upgraded from fc6 x86_64 too. Again, I can't run various things that use PolicyKit, including gnome-system-monitor, polkit-gnome-example and gpk-update-icon. A few lines from an strace of polikit-gnome-example that also occur when the other apps are run: inotify_init() = 6 inotify_add_watch(6, "/etc/PolicyKit/PolicyKit.conf", IN_MODIFY|IN_ATTRIB|IN_CREATE) = -1 ENOSPC (No space left on device) write(2, "[WARN 20006] polkit-error.c:143:"..., 80[WARN 20006] polkit-error.c:143:polkit_error_get_error_message(): error != NULL ) = 80 write(2, " Not built with -rdynamic so una"..., 57 Not built with -rdynamic so unable to print a backtrace ) = 57 write(2, "\n** (polkit-gnome-example:20006)"..., 93 ** (polkit-gnome-example:20006): WARNING **: Failed to initialize PolicyKit context: (null) (and, no, my disc _isn't_ full) --- Additional comment from john@redux.org.uk on 2008-10-15 12:50:09 EDT --- Oh, and why is this severity "low"? An important part of the system stops working and the only way is to fix it is to log out of your whole desktop session? Where "low" is meant to be: Problem is minor loss of function, or other problem where easy workaround is present. Not severity "low" in my mind. --- Additional comment from rhughes@redhat.com on 2008-10-15 13:58:41 EDT --- (In reply to comment #8) > Oh, and why is this severity "low"? Because it's low priority for me, at the moment. Does this happen on a fresh install, or just from upgrades? What is the difference in packages installed from an F8->F9 upgrade and a fresh F9 install? Richard. --- Additional comment from aravind.parchuri@gmail.com on 2008-10-15 15:47:30 EDT --- I found the bug on a fresh F9 install. --- Additional comment from john@redux.org.uk on 2008-10-15 17:47:15 EDT --- On my upgraded an fully updated system: # rpm -qa|grep -i policy PolicyKit-gnome-0.8-4.fc9.x86_64 PolicyKit-gnome-libs-0.8-4.fc9.x86_64 PolicyKit-gnome-demo-0.8-4.fc9.x86_64 selinux-policy-3.3.1-95.fc9.noarch PolicyKit-0.8-3.fc9.x86_64 selinux-policy-targeted-3.3.1-95.fc9.noarch policycoreutils-2.0.52-8.fc9.x86_64 checkpolicy-2.0.16-3.fc9.x86_64 # rpm -qa|grep -i dbus ndesk-dbus-glib-0.4.1-3.fc9.x86_64 dbus-glib-devel-0.74-8.fc9.x86_64 dbus-debuginfo-1.2.1-1.fc9.x86_64 dbus-libs-1.2.4-1.fc9.x86_64 dbus-glib-0.74-8.fc9.i386 dbus-1.2.4-1.fc9.x86_64 dbus-devel-1.2.4-1.fc9.x86_64 dbus-glib-debuginfo-0.74-8.fc9.x86_64 dbus-libs-1.2.4-1.fc9.i386 dbus-sharp-0.63-9.fc9.x86_64 dbus-glib-0.74-8.fc9.x86_64 dbus-python-0.83.0-2.fc9.x86_64 dbus-x11-1.2.4-1.fc9.x86_64 dbus-qt-0.70-4.fc9.x86_64 ndesk-dbus-0.6.1a-2.fc9.x86_64 I was about to paste a list of packages on a clean install that had been fully updated, but I tried to get such a system up to date, and got to a point where gpk-update-icon stopped working in a different way, namely: process 3394: arguments to dbus_message_new_method_call() were incorrect, assertion "_dbus_check_is_valid_path (path)" failed in file dbus-message.c line 1074. This is normally a bug in some application using the D-Bus library. D-Bus not built with -rdynamic so unable to print a backtrace At this point: # rpm -qa |grep -i policy PolicyKit-0.8-2.fc9.x86_64 checkpolicy-2.0.14-1.fc9.x86_64 policycoreutils-2.0.46-5.fc9.x86_64 selinux-policy-targeted-3.3.1-42.fc9.noarch PolicyKit-gnome-libs-0.8-4.fc9.x86_64 PolicyKit-gnome-0.8-4.fc9.x86_64 policycoreutils-gui-2.0.46-5.fc9.x86_64 selinux-policy-3.3.1-42.fc9.noarch # rpm -qa |grep -i dbus dbus-libs-1.2.1-1.fc9.x86_64 dbus-1.2.1-1.fc9.x86_64 dbus-x11-1.2.1-1.fc9.x86_64 ndesk-dbus-glib-0.4.1-3.fc9.x86_64 dbus-glib-0.74-6.fc9.x86_64 dbus-python-0.82.4-2.fc9.x86_64 dbus-qt-0.70-4.fc9.x86_64 ndesk-dbus-0.6.1a-2.fc9.x86_64 After a 'yum upgrade' : # rpm -qa|grep -i policy PolicyKit-0.8-3.fc9.x86_64 checkpolicy-2.0.16-3.fc9.x86_64 PolicyKit-gnome-0.8-4.fc9.x86_64 selinux-policy-targeted-3.3.1-95.fc9.noarch selinux-policy-3.3.1-95.fc9.noarch selinux-policy-devel-3.3.1-95.fc9.noarch policycoreutils-gui-2.0.52-8.fc9.x86_64 PolicyKit-gnome-libs-0.8-4.fc9.x86_64 policycoreutils-2.0.52-8.fc9.x86_64 # rpm -qa|grep -i dbus dbus-x11-1.2.4-1.fc9.x86_64 dbus-glib-0.74-8.fc9.x86_64 dbus-qt-0.70-4.fc9.x86_64 ndesk-dbus-0.6.1a-2.fc9.x86_64 dbus-1.2.4-1.fc9.x86_64 dbus-python-0.83.0-2.fc9.x86_64 ndesk-dbus-glib-0.4.1-3.fc9.x86_64 dbus-libs-1.2.4-1.fc9.x86_64 Note, the clean installation is only a test system, I've not used it enough to say whether or not this problem will occur or not. I can provide the complete 'rpm -qa' for either of these systems if required. --- Additional comment from dennis.wagelaar@vub.ac.be on 2008-10-16 04:15:35 EDT --- For me, this happened on two x86_64 boxes. Both of them were upgraded via yum from F8 to F9 conform http://fedoraproject.org/wiki/YumUpgradeFaq current package versions: - PolicyKit-0.8-3.fc9.x86_64 - PolicyKit-gnome-0.8-4.fc9.x86_64 N.B. I no longer have the i386 version of PolicyKit installed. I also no longer manage to reproduce the problem. I ran the following command to clean up my installed packages: # package-cleanup --cleandupes Can you guys check if this works for you too? --- Additional comment from john@redux.org.uk on 2008-10-17 04:37:50 EDT --- (In reply to comment #12) [snip] > N.B. I no longer have the i386 version of PolicyKit installed. I also no longer > manage to reproduce the problem. I ran the following command to clean up my > installed packages: > > # package-cleanup --cleandupes > > Can you guys check if this works for you too? # package-cleanup --cleandupes Setting up yum No dupes to clean --- Additional comment from rhughes@redhat.com on 2008-10-27 09:12:23 EDT --- *** Bug 452080 has been marked as a duplicate of this bug. *** --- Additional comment from scott@sonic.net on 2008-11-08 22:30:58 EDT --- Created an attachment (id=322987) typescript of gdb session, attempting to run gpk-update-viewer I have this same problem with rawhide (f10) -- attached is typescript from script(1) showing multiple segfaults w/their backtraces... I first noticed this problem when I could not spawn the update viewer from the update applet. If it matters, this is a gnome desktop, using compiz (spawned with compiz-manager --replace) and emerald (emerald --replace).
the strace several comments up looks like something might be eating up all available inotify watches. Are you running some inotify hog, like trackerd ?
Created attachment 323037 [details] ps -axww output re: me running something hogging up inotify watches None that I know of -- I've attached a ps output. I was also running ccsm, and had the compiz filesystem enabled, but shutting it off didn't clear up the problem. -Scott
inotify hog could be beagle, possibly.
Good call! It was beagle on my system. As soon as I stopped beagle, the previously crashing applications started working properly again. I have around 43k directories/sub-directories in my home directory, and I guess beagle was using up all 8192 of the available inotify watches. I also verified that it was the inotify watch limit by doing an "inotifywatch -r" to set up around 8k watches and managed to reproduce the crash. Any plans to increase the default fs.inotify.max_user_watches ?
Aha, killing beagled fixes the problems for me too!
I haven't tried this yet, as I'm at the office, but I shall this afternoon. This is very cool! And it makes sense: iirc, i remember seeing that beagled has a delay that it waits at startup before it starts doing its stuff. I wonder how SuSE handles this situation? -Scott
This bug appears to have been reported against 'rawhide' during the Fedora 10 development cycle. Changing version to '10'. More information and reason for this action is here: http://fedoraproject.org/wiki/BugZappers/HouseKeeping
Just to confirm this happens with FC10 as well. Right clicking on the desktop and then Change Desktop Background would crash. Killing beagled resolved this.
When is this going to get fixed? I just ran into this on my FC10 x86_64 system. I only figured it out after using strace and google.
This message is a reminder that Fedora 10 is nearing its end of life. Approximately 30 (thirty) days from now Fedora will stop maintaining and issuing updates for Fedora 10. It is Fedora's policy to close all bug reports from releases that are no longer maintained. At that time this bug will be closed as WONTFIX if it remains open with a Fedora 'version' of '10'. Package Maintainer: If you wish for this bug to remain open because you plan to fix it in a currently maintained version, simply change the 'version' to a later Fedora version prior to Fedora 10's end of life. Bug Reporter: Thank you for reporting this issue and we are sorry that we may not be able to fix it before Fedora 10 is end of life. If you would still like to see this bug fixed and are able to reproduce it against a later version of Fedora please change the 'version' of this bug to the applicable version. If you are unable to change the version, please add a comment here and someone will do it for you. Although we aim to fix as many bugs as possible during every release's lifetime, sometimes those efforts are overtaken by events. Often a more recent Fedora release includes newer upstream software that fixes bugs or makes them obsolete. The process we are following is described here: http://fedoraproject.org/wiki/BugZappers/HouseKeeping
Fedora 10 changed to end-of-life (EOL) status on 2009-12-17. Fedora 10 is no longer maintained, which means that it will not receive any further security or bug fix updates. As a result we are closing this bug. If you can reproduce this bug against a currently maintained version of Fedora please feel free to reopen this bug against that version. Thank you for reporting this bug and we are sorry it could not be fixed.