[Userland-team] [OpenIndiana Distribution - Bug #1987] dbus taking 100% of cpu
illumos project
devnull at illumos.org
Sun Apr 29 06:47:23 UTC 2012
Issue #1987 has been updated by Richard PALO.
In trying to debug https://www.illumos.org/issues/2653#change-6939 I'm experiencing this as one of the first steps in the problem evolution. dbus-daemon, and a number of other session processes from a logged out VNC user's session. (note VNC is in multiuser, *inetd/wait=false*)
here are the last few lines from dapptrace -ap <dbus-daemon-pid>
<pre>
5904/1: 41992242 . . -> dbus-daemon:_dbus_poll(0x80474C0, 0x2, 0xFFFFFFFF)
5904/1: 41992243 . . -> libc.so.1:poll(0x80474C0, 0x2, 0xFFFFFFFF)
5904/1: 41992243 . . -> libc.so.1:_pollsys(0x80474C0, 0x2, 0x0)
5904/1: 41992244 . . -> libc.so.1:__pollsys(0x80474C0, 0x2, 0x0)
5904/1: 41992247 13 3 <- libc.so.1:__pollsys = 13
5904/1: 41992248 37 4 <- libc.so.1:_pollsys = 423
5904/1: 41992248 58 5 <- libc.so.1:poll = 83
5904/1: 41992249 80 6 <- dbus-daemon:_dbus_poll = 47
5904/1: 41992249 . . -> dbus-daemon:check_babysit_events(0x1711, 0x10, 0x1)
5904/1: 41992250 . . -> libc.so.1:waitpid(0x1711, 0x8047454, 0x40)
5904/1: 41992250 . . -> libc.so.1:waitid(0x0, 0x1711, 0x80473B0)
5904/1: 41992251 . . -> libc.so.1:__waitid(0x0, 0x1711, 0x80473B0)
5904/1: 41992253 12 2 <- libc.so.1:__waitid = 21
5904/1: 41992253 34 3 <- libc.so.1:waitid = 274
5904/1: 41992254 . . -> libc.so.1:wstat(0x0, 0x0, 0x80473B0)
5904/1: 41992254 9 0 <- libc.so.1:wstat = 79
5904/1: 41992255 75 5 <- libc.so.1:waitpid = 149
5904/1: 41992255 103 6 <- dbus-daemon:check_babysit_events = 177
5904/1: 41992256 . . -> dbus-daemon:_dbus_poll(0x80474C0, 0x2, 0xFFFFFFFF)
5904/1: 41992256 . . -> libc.so.1:poll(0x80474C0, 0x2, 0xFFFFFFFF)
5904/1: 41992257 . . -> libc.so.1:_pollsys(0x80474C0, 0x2, 0x0)
dtrace: 523021 drops on CPU 7
CALL COUNT
libc.so.1 __pollsys 1108
dbus-daemon _dbus_poll 1110
dbus-daemon check_babysit_events 1110
libc.so.1 __waitid 1110
libc.so.1 _pollsys 1110
libc.so.1 poll 1110
libc.so.1 waitid 1110
libc.so.1 waitpid 1110
libc.so.1 wstat 1111
</pre>
I'm trying now to collect information on another step in the evolution, where xvnc goes pegs out at 100% cpu.
FWIW, the other processes hanging around are:
<pre>
# ps -u <user>
PID TTY TIME CMD
5771 ? 0:00 gvfsd-me
5764 ? 0:00 gvfs-hal
5759 ? 0:00 gvfsd-tr
5751 ? 0:00 gvfsd
5904 ? 871:32 dbus-dae
5905 ? 0:00 gconfd-2
</pre>
----------------------------------------
Bug #1987: dbus taking 100% of cpu
https://www.illumos.org/issues/1987
Author: Jeffrey Baitis
Status: New
Priority: Normal
Assignee: OI Userland
Category:
Target version: oi_151_stable
Difficulty: Medium
Tags: dbus
After my 151a machine had about 39 days of uptime with heavy X11 use, top noticed that dbus-daemon was taking 100% of the cpu time in one of my CPU cores. svcadm disable dbus did not seem to kill the rogue dbus-daemon; I had to kill it by pid manually.
I ran the Proc/dapptrace script from the DTrace toolkit. Here's the output that I got:
<pre>
CALL(args) = return
-> check_babysit_events(0x2CF8, 0x20, 0x1)
<- check_babysit_events = 177
-> _dbus_poll(0x8047490, 0x2, 0xFFFFFFFF)
<- _dbus_poll = 47
-> check_babysit_events(0x2CF8, 0x20, 0x1)
<- check_babysit_events = 177
-> _dbus_poll(0x8047490, 0x2, 0xFFFFFFFF)
<- _dbus_poll = 47
-> check_babysit_events(0x2CF8, 0x20, 0x1)
<- check_babysit_events = 177
-> _dbus_poll(0x8047490, 0x2, 0xFFFFFFFF)
<- _dbus_poll = 47
-> check_babysit_events(0x2CF8, 0x20, 0x1)
<- check_babysit_events = 177
-> _dbus_poll(0x8047490, 0x2, 0xFFFFFFFF)
<- _dbus_poll = 47
-> check_babysit_events(0x2CF8, 0x20, 0x1)
<- check_babysit_events = 177
-> _dbus_poll(0x8047490, 0x2, 0xFFFFFFFF)
<- _dbus_poll = 47
</pre>
... ad nauseam
--
You have received this notification because you have either subscribed to it, or are involved in it.
To change your notification preferences, please click here: http://www.illumos.org/my/account
More information about the Userland-team
mailing list