[MASTER] E-D-S hangs on login and uses 100% cpu

Bug #151536 reported by Michael Vogt on 2007-10-11
228
This bug affects 17 people
Affects Status Importance Assigned to Milestone
evolution-data-server
Expired
Critical
evolution-data-server (Ubuntu)
High
Unassigned

Bug Description

Binary package hint: evolution-data-server

<PLEASE> -- do not add simple "me too" comments, they do not help: we already *know* this is a problem, and these comments only make it more difficult to find potentially important data in the bug. If you have the knowledge and time to help debug, your help *is* appreciated. Go to #ubuntu-bugs in irc.freenode.net, and discuss it there. I am sure that any one of us will be more than happy to get on with this ;-)</PLEASE>

On login evolution-data-server uses 100% cpu. At login evolution-alarm-notify is auto-started by gnome-session; e-a-n kicks in E-D-S and, for yet an unknown reason, E-D-S enters a loop. strace on E-D-S shows a sequence of very fast poll() calls; a strace on e-a-n shows e-a-n waiting for an event. This seems to happen on some machines, but not all. It affects both Evolution users and non-users -- E-D-S is still installed under Gnome even if you do not use the Evolution mail client.

All symptoms are of a race condition, very early in E-D-S or a library startup. Backtraces are inconclusive, and show slightly different code paths. I cannot reproduce it on my machine (an AMD64 dual-core), no matter what I try.

I would like to know what Linux and hardware architecture, and how many CPUs for those that are *frequently* hit (for example, {linux64, AMD64, 2 CPUs}, meaning I am running linux64 on an AMD64 with two CPUs; {linux32, AMD64, 4 CPUs}, meaning I am running linux32 Intel on an AMD64 with 4 CPUs).

Right now we are considering this as a race very early on session startup, perhaps caused by the current placement of e-a-n on the default session list (with the default priority of 50). I myself still cannot reproduce it, and I wonder if having a dual-CPU machine, and/or linux64, and/or AMD64, makes the race more difficult to happen.

Michael Vogt (mvo) wrote :

(gdb) thread apply all bt full

Thread 3 (Thread -1224352880 (LWP 6553)):
#0 0xffffe410 in __kernel_vsyscall ()
No symbol table info available.
#1 0xb76ec647 in poll () from /lib/tls/i686/cmov/libc.so.6
No symbol table info available.
#2 0xb77f5593 in ?? () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#3 0x080c37a8 in ?? ()
No symbol table info available.
#4 0x00000009 in ?? ()
No symbol table info available.
#5 0xffffffff in ?? ()
No symbol table info available.
#6 0x080c37a8 in ?? ()
No symbol table info available.
#7 0x0000000b in ?? ()
No symbol table info available.
#8 0xb776f2d6 in ?? () from /lib/tls/i686/cmov/libpthread.so.0
No symbol table info available.
#9 0xb7774541 in pthread_mutex_lock () from /lib/tls/i686/cmov/libpthread.so.0
No symbol table info available.
#10 0xb77f5909 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#11 0xb7909c20 in ?? () from /usr/lib/libORBit-2.so.0
No symbol table info available.
#12 0x080531c8 in ?? ()
No symbol table info available.
#13 0xb787f21c in ?? () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#14 0xb705d3d8 in ?? ()
No symbol table info available.
#15 0xb78155af in ?? () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#16 0x00000000 in ?? ()
No symbol table info available.

Thread 2 (Thread -1260389488 (LWP 6574)):
#0 0xffffe410 in __kernel_vsyscall ()
No symbol table info available.
#1 0xb76ec647 in poll () from /lib/tls/i686/cmov/libc.so.6
No symbol table info available.
#2 0xb77f5593 in ?? () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
---Type <return> to continue, or q <return> to quit---
#3 0x080815c8 in ?? ()
No symbol table info available.
#4 0x00000001 in ?? ()
No symbol table info available.
#5 0xffffffff in ?? ()
No symbol table info available.
#6 0x080815c8 in ?? ()
No symbol table info available.
#7 0x00000001 in ?? ()
No symbol table info available.
#8 0xb776a140 in ?? () from /lib/tls/i686/cmov/libc.so.6
No symbol table info available.
#9 0xb7774541 in pthread_mutex_lock () from /lib/tls/i686/cmov/libpthread.so.0
No symbol table info available.
#10 0xb77f5909 in g_main_loop_run () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#11 0xb7f50932 in ?? () from /usr/lib/libebook-1.2.so.9
No symbol table info available.
#12 0x08081168 in ?? ()
No symbol table info available.
#13 0x00000000 in ?? ()
No symbol table info available.

Thread 1 (Thread -1221720400 (LWP 6522)):
#0 0xb77f04d4 in ?? () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#1 0xbfa97e48 in ?? ()
No symbol table info available.
#2 0xb77f20c3 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
Backtrace stopped: frame did not save the PC
#0 0xb77f04d4 in ?? () from /usr/lib/libglib-2.0.so.0
(gdb)

Markus Kienast (elias1884) wrote :

As far as evolution-data-server consuming 100% cpu on login is concerned, I can confirm this. Surprisingly the load gets down to normal, when I start evolution.

C de-Avillez (hggdh2) wrote :

Thanks, mvo and Elias, for the bug report (and, of course, we are sorry it took us that long to get to it).

Unfortunately, the backtrace is incomplete -- there is a suggestion we are looping trying to acquire a lock, but no real data on who is doing that. Could you please repeat it with debug symbols for (at least) libglib and e-d-s?

Question: was e-d-s already running on login (in other words, did you log out Gnome and log in again)? Does it happen when you log out, or only during login? Is Evo itself auto-started on login?

Changed in evolution-data-server:
assignee: nobody → desktop-bugs
importance: Undecided → Medium
status: New → Incomplete

This happens with first login after boot up.

What is started automatically is the evolution alarm notifier.

Other than that there is to mention that I have a personal and a
business calendar. I had also a Birthdays and Anniversaries calendar but
deactivated due to comments on another e-d-s related bug report, that
this type of calendar causes evolution to not exit cleanly on logout.

Elias, I am now unsure on your issue: the bug relates to evolution-data-server apparent loop, but you state (or I understood so) it was evolution-alarm-notify looping. Can you please clarify?

I am also concerned by this bug.

I found that it occurs when I add a Web Calendar (Google Calendar to be precise), and reboot the laptop. Everytime I log into Gnome, the evolution-data-server take 100% cpu and I must do evolution --force-shutdown to stop it.

I have not see this bug without a web calendar configured.

Michael Vogt (mvo) wrote :

Here is a updated backtrace:

$ gdb -p 6969
(gdb) thread apply all bt

Thread 3 (Thread 0xb6f95b90 (LWP 7016)):
#0 0xffffe410 in __kernel_vsyscall ()
#1 0xb7634697 in poll () from /lib/tls/i686/cmov/libc.so.6
#2 0xb7749166 in g_main_context_iterate (context=0x8064d90, block=1,
    dispatch=1, self=0x8061128)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:3006
#3 0xb7749517 in IA__g_main_loop_run (loop=0x80531b0)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:2905
#4 0xb7848c30 in ?? () from /usr/lib/libORBit-2.so.0
#5 0x080531b0 in ?? ()
#6 0xb77bc248 in ?? () from /usr/lib/libglib-2.0.so.0
#7 0xb6f953d8 in ?? ()
#8 0xb776e01f in g_thread_create_proxy (data=0x8061128)
    at /build/buildd/glib2.0-2.15.3/glib/gthread.c:635
Backtrace stopped: frame did not save the PC

Thread 2 (Thread 0xb4dffb90 (LWP 7042)):
#0 0xffffe410 in __kernel_vsyscall ()
#1 0xb7634697 in poll () from /lib/tls/i686/cmov/libc.so.6
#2 0xb7749166 in g_main_context_iterate (context=0x80894c8, block=1,
    dispatch=1, self=0x8082868)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:3006
#3 0xb7749517 in IA__g_main_loop_run (loop=0x8089348)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:2905
#4 0xb7f8c592 in ?? () from /usr/lib/libebook-1.2.so.9
#5 0x08089348 in ?? ()
#6 0x00000000 in ?? ()

Thread 1 (Thread 0xb723f8c0 (LWP 6969)):
#0 0xb76be595 in pthread_mutex_lock () from /lib/tls/i686/cmov/libpthread.so.0
#1 0xb7745dd4 in IA__g_main_context_dispatch (context=0x805d2a8)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:2074
#2 0xb7749133 in g_main_context_iterate (context=0x805d2a8, block=1,
    dispatch=1, self=0x804e490)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:2697
#3 0xb7749517 in IA__g_main_loop_run (loop=0x80a9330)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:2905
#4 0xb7895d53 in bonobo_main () from /usr/lib/libbonobo-2.so.0
#5 0x0804b974 in ?? ()
#6 0x0804c3c7 in ?? ()
#7 0x00000020 in ?? ()
---Type <return> to continue, or q <return> to quit---
#8 0x0804c4e3 in ?? ()
#9 0x00000000 in ?? ()
#0 0xb76be595 in pthread_mutex_lock () from /lib/tls/i686/cmov/libpthread.so.0

$ strace -p 6969
poll([{fd=5, events=POLLIN}, {fd=7, events=POLLIN|POLLPRI}], 2, 0) = 0

$ ltrace -p 6969
--- SIGSTOP (Stopped (signal)) ---
--- SIGSTOP (Stopped (signal)) ---

Michael Vogt (mvo) wrote :

>Question: was e-d-s already running on login (in other words, did you log out Gnome >and log in again)? Does it happen when you log out, or only during login? Is Evo >itself auto-started on login?

This happens on a fresh boot, when I login the first time into gnome (this is with the current hardy now). When I kill evolution-data-server and logout/login then it does not happen anymore. Evo does not start for me (I don't use it at all).

Cheers,
 Michael

Changed in evolution-data-server:
status: Incomplete → New
C de-Avillez (hggdh2) wrote :

Thank you, Michael. Unfortunately, the stacktrace is still incomplete -- Although it looks like we are looping in pthread_mutex_lock (which calls __pthread_lock, which may go into a spinloop), we still do not know which e-d-s module initiated the call, nor when. Could you please install e-d-s-dbgsym (in fact, please install all corresponding e-d-s dbgsym packages, see below for a list of potential packages) )and repeat the bt with a 'thread apply all bt full'?

This is still very weird... a loop at this point suggests something else is broken at a basic level... or some structures are being overlaid.

E-D-S builds the following packages; some of them will be installed, please install the corresponding dbgsym packages (and please adjust version info as needed):

evolution-data-server
 libcamel1.2-10
 libebook1.2-9
 libecal1.2-7
 libedata-book1.2-2
 libedata-cal1.2-6
libedataserver1.2-9
 libedataserverui1.2-8
 libegroupwise1.2-13
 libexchange-storage1.2-3
 libgdata-google1.2-1
 libgdata1.2-1

Changed in evolution-data-server:
status: New → Incomplete
C de-Avillez (hggdh2) wrote :

a brief chat on #evolution suggests this might be related to having no network at login (one of them has seen something like this before).

Yeah, I think it's the problem because if you use NetworkManager, the connection is not ready when e-d-s start.

So, I think that if you have a web calendar and the connection is not ready, e-d-s start like crazy...

C de-Avillez (hggdh2) wrote :

I tried without network:

(1) standard login after boot, with a monitor running on tty1. No loops.

(2) disabling NetworkManager, and disconnecting from the wired network; rebooted and logged in, with a monitor running on tty1. No loops.

So I cannot repeat the issue. I depend on you to get a backtrace with E-D-S symbols so that we can dig further on this.

Michael Vogt (mvo) wrote :
Download full text (3.8 KiB)

Here is a new backtrace:

(gdb) thread apply all bt full

Thread 3 (Thread 0xb6ea0b90 (LWP 7106)):
#0 0xffffe410 in __kernel_vsyscall ()
No symbol table info available.
#1 0xb753c697 in poll () from /lib/tls/i686/cmov/libc.so.6
No symbol table info available.
#2 0xb7651166 in g_main_context_iterate (context=0x8064df0, block=1,
    dispatch=1, self=0x8060e10)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:3006
 got_ownership = <value optimized out>
 max_priority = 2147483647
 timeout = -1
 some_ready = <value optimized out>
 nfds = 11
 allocated_nfds = <value optimized out>
 fds = (GPollFD *) 0x80bdcb0
 __PRETTY_FUNCTION__ = "g_main_context_iterate"
#3 0xb7651517 in IA__g_main_loop_run (loop=0x80531b0)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:2905
 got_ownership = -1218681568
 self = (GThread *) 0x8060e10
 __PRETTY_FUNCTION__ = "IA__g_main_loop_run"
#4 0xb7750c30 in ?? () from /usr/lib/libORBit-2.so.0
No symbol table info available.
#5 0x080531b0 in ?? ()
No symbol table info available.
#6 0xb76c4248 in ?? () from /usr/lib/libglib-2.0.so.0
No symbol table info available.
#7 0xb6ea03d8 in ?? ()
No symbol table info available.
#8 0xb767601f in g_thread_create_proxy (data=0x8060e10)
    at /build/buildd/glib2.0-2.15.3/glib/gthread.c:635
 __PRETTY_FUNCTION__ = "g_thread_create_proxy"
Backtrace stopped: frame did not save the PC

Thread 2 (Thread 0xb4cfeb90 (LWP 7126)):
#0 0xffffe410 in __kernel_vsyscall ()
No symbol table info available.
#1 0xb753c697 in poll () from /lib/tls/i686/cmov/libc.so.6
No symbol table info available.
#2 0xb7651166 in g_main_context_iterate (context=0x807fb68, block=1,
    dispatch=1, self=0x8072900)
---Type <return> to continue, or q <return> to quit---
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:3006
 got_ownership = <value optimized out>
 max_priority = 2147483647
 timeout = -1
 some_ready = <value optimized out>
 nfds = 1
 allocated_nfds = <value optimized out>
 fds = (GPollFD *) 0x8084598
 __PRETTY_FUNCTION__ = "g_main_context_iterate"
#3 0xb7651517 in IA__g_main_loop_run (loop=0x8078070)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:2905
 got_ownership = -1218681568
 self = (GThread *) 0x8072900
 __PRETTY_FUNCTION__ = "IA__g_main_loop_run"
#4 0xb7e94592 in startup_mainloop (arg=0x0) at e-book.c:3765
 loop = (GMainLoop *) 0xfffffffc
#5 0xb767601f in g_thread_create_proxy (data=0x8072900)
    at /build/buildd/glib2.0-2.15.3/glib/gthread.c:635
 __PRETTY_FUNCTION__ = "g_thread_create_proxy"
#6 0xb75c44fb in start_thread () from /lib/tls/i686/cmov/libpthread.so.0
No symbol table info available.
#7 0xb75468ee in clone () from /lib/tls/i686/cmov/libc.so.6
No symbol table info available.

Thread 1 (Thread 0xb71478c0 (LWP 7084)):
#0 0xb7650db6 in g_main_context_iterate (context=0x805d2a8, block=1,
    dispatch=1, self=0x804e490)
    at /build/buildd/glib2.0-2.15.3/glib/gmain.c:2666
 max_priority = <value optimized out>
 timeout = <value optimized out>
 some_ready = <value optimized out>
 nfds = <value optimized out>
 allocated_nfds = <value optimized out>
 fds = <value optimized out>
 __PRETTY_FUNCTION__ = "g_main_context_iterate"
#1 0xb7651517 in IA__g_main_loop_run (loop=0x808e038)...

Read more...

Anton Kudris (kudris) wrote :

I have the same bug. If anyone can tell me how to create stacktrace I could try to help as well

C de-Avillez (hggdh2) wrote :

@Anton Kudris: after you find yourself in such a situation, find out the pid of the offending process, and run:

gdb -p <pid> # where <pid> is the pid you found above

once in gdb, issue "thread apply all bt full" -- this will dump the backtraces for all threads.

So far I have been unable to reproduce it, or to make sense of the stacktraces we have collected so far. Perhaps it would help to go into some cycles of (interrupt, dump stacks, continue), to see if we can dig into whatever is causing this. Still, this seems to happen very early in e-d-s startup.

No matter what, the backtraces will only have value if the debug packages (either the dbgsym or dbg) for e-d-s (*all* of them), glib, libc6, and libbonobo (at least) are installed.

A Kao (ak-ubuntu) wrote :

This also affects a clean install of Hardy Heron Alpha 5.

Caroline Ford (secretlondon) wrote :

Without wishing to do a "me too" I also have in Hardy. I've set up a web calendar if that helps.

Caroline Ford (secretlondon) wrote :

This is from gdb but it stopped using 100% cpu after I startred gbd.

The versions of syndbg packages were too out of date to install so not sure how useful this is.

Caroline Ford (secretlondon) wrote :
Caroline Ford (secretlondon) wrote :

eds stops using 100% cpu when gdb is attached to it, and resumes when it is closed. I am told this isn't interesting :)

I have a better trace from gdb - but for some reason it isn't picking up libc symbols. I will try and reboot and see if it fixes it.

Caroline Ford (secretlondon) wrote :

First reboot it didn't happen. I loaded evolution itself to see if it would trigger it, it didn't but then my laptop froze hard (this may be random).

Second reboot it did it. It's still not picking up libc symbols.

I don't have internet access when eds starts - nor do I use network manager. This means that evo and firefox 3 now start in offline mode.

Caroline Ford (secretlondon) wrote :

I ran strace on eds for a few seconds.

Process 5690 attached with 3 threads - interrupt to quit
[pid 5679] 20:03:38.365141 restart_syscall(<... resuming interrupted call ...> <unfinished ...>
[pid 5690] 20:03:38.365483 restart_syscall(<... resuming interrupted call ...> <unfinished ...>
[pid 5665] 20:03:38.365589 poll([{fd=5, events=POLLIN}, {fd=7, events=POLLIN|POLLPRI}], 2, 0) = 0
[pid 5665] 20:03:38.365698 poll([{fd=5, events=POLLIN}, {fd=7, events=POLLIN|POLLPRI}], 2, 0) = 0
....
5665 is eds. The strace output is all the same thing, repeated many times a second. This may or may not be interesting.

C de-Avillez (hggdh2) wrote :

Caroline's bt is good enough; also the strace clearly shows very fast polls. Opened https://bugs.launchpad.net/ubuntu/+source/evolution-data-server/+bug/151536 upstream. Finally!

Changed in evolution-data-server:
status: Incomplete → Triaged
C de-Avillez (hggdh2) wrote :

Ugh! what time zone differences made me do? Of course, the upstream bug is not on LP, but gnome: http://bugzilla.gnome.org/show_bug.cgi?id=518524

Changed in evolution-data-server:
status: Unknown → New
C de-Avillez (hggdh2) wrote :

@ Caroline (or any other able to reproduce this): could you please repeat the issue, and get gdb backtraces (thread apply all bt full) from *both* e-d-s and evolution-alarm-notify? Upstream is, right now, as mystified as we are...

C de-Avillez (hggdh2) wrote :

also -- an 'ifconfig -a' would be interesting to see.

I'll be back at home early evening UTC and will see what I can manage.

Caroline Ford (secretlondon) wrote :

I think this started yesterday as a result of me doing one of two
things - either installing evolution itself (it had got uninstalled as
a result of a dist-update) or setting up my google calendar to sync
with evo. I fed evo false email info to allow me to do that..

It's possible that I don't have all of evo installed that I should
have. I don't have exchange server installed, for example.

This may be random and/or expected but with eds frozen by gdb when I clicked on the clock the entire gnome panel died. Unattaching gdb from eds got me the panel back..

Talk on #ubuntu-bugs suggests that clicking on the clock and the panel going away has happened to others without having gdb involved.

Jeffrey Baker (jwbaker) wrote :

I don't think this has anything to do with web calendars because I haven't touched evolution at all (not even launched it a single time) and I still get this 100% looping behavior at the first login after boot. I believe the trigger is no network at login time.

I get it most (75%+) of the time. The workround is to find the process
using top and then attach gdb to it!

This means you can't use evolution, but it gives you your computer back..

C de-Avillez (hggdh2) wrote :

> I get it most (75%+) of the time. The workround is to find the process
> using top and then attach gdb to it!

> This means you can't use evolution, but it gives you your computer
> back..

Or to issue (from a terminal in the X session):

evolution --force-shutdown

Or to simply (not really that different) kill the offending process.

The weirdest of it all is that most people have this behaviour once in
a long while, and Caroline has it most of the time (and, as she told me,
while she was already connected to a PPP session). Hum. Network-Manager
does not really work or recognise PPP as a network connection. Hum.

It may well be that something related to network, or its absence, plays
a role here, but we have been unable to identify what/how.

All I can see from the backtraces is that E-D-S itself is looping *very*
early in the E-D-S startup... so early that I wonder if this is a Evo
issue, or Gnome.

Anyway, once you stop it, everybody (so far) reports it can be restarted
without being hit by the loop.

Caroline Ford (secretlondon) wrote :

Well if I unattach gdb it will go back to 100% cpu usage. I appreciate
that probably isn't the same as stopping it.

There is a bug about firefox and evo about the fact that Firefox
thinks we are offline when we use ppp connections.

nm does respond in syslog when I make my connection though.

Caroline Ford (secretlondon) wrote :

kill -9 process did actually get rid of it.

Mar 2 02:41:02 celery NetworkManager: <debug> [1204425662.393480]
nm_hal_device_added(): New device added (hal udi is
'/org/freedesktop/Hal/devices/net_computer_0').

The above does appear in syslog during wvdial's ppp connection.

Confirmed it is bug. evolution-data-server consume 100% CPU too in my PC. Only happens after the last two updates I did within the past week.

Using Ubuntu Hardy Alpha 5.

Tomas M. (el-dragon) wrote :

ive got the same problem with evolution-data-server.

under hardy alpha6.

i had the problem with firefox thinking it was offline, but ive fixed it adding the following to /etc/network/interfaces:
iface wlan0 inet dhcp
wireless-essid some_ESSID
auto wlan0

now it doesnt think its offline anymore. but the issue concerning e-d-s remains.

only difference between hardy and gutsy is i didnt have an eth0 available at the time, now i have (details concerning 8139too under PIO mode)

eth0 is set on roaming mode (aka no entry under /etc/network/interfaces )

dont know if this helps much.

will try later to blacklist my ethernet module to see if it happens

Jim Qode (jimqode) wrote :

I confirm this bug under Hardy Alpha 6.

ergo (ergo14) wrote :

well one of my pc doesnt hang on login, but evolution-data-(server i guess) process is topping 100% of cpu power, when i do evolution --force-shutdown, everything gets back to normal.
im on 8.04beta. Maybe it has something to do with wireless card ?

ynamestnikov (ynamestnikov) wrote :

to Ergo: I don't use evolution and I haven't got any wireless card, but after login evolution data server + trackerd use 100% of cpu. Strange isn't it?)

Hardy Beta.

C de-Avillez (hggdh2) on 2008-04-13
Changed in evolution-data-server:
importance: Medium → High
C de-Avillez (hggdh2) on 2008-04-26
description: updated
Changed in evolution-data-server:
status: New → Confirmed
81 comments hidden view all 161 comments
Alterac (arkaine-pk-12) wrote :

I just updated to Hardy 8.04 about a few days ago. Before, on Gutsy 7.10, I did not have this problem.

I am using a Dual Core Laptop and the process only consumes 50% of my processor, that is, 100% of each core at a single time.

I am running linux32, Intel Core2 Duo Mobile Processor T7400, 2 CPUs.

mawg (dogintub) wrote :

os: Ubuntu 8.04 i386 (upgraded from 7.10)
cpu: 2x AMD Athlon MP 1.5ghz
motherboard: MSI-6501 (K7D-L)
using gnome

Sebastien Bacher (seb128) wrote :

does disabling the autostart rather than changing the priority workaround the issue?

Andrew (andrewkk) wrote :

{linux32, AMD64, 2 CPUs} (Intel Core 2 Duo)

Might just be a coincidence, but I'd never seen this happen before until I started switching between users via GDM. Previously I had only one user account so I'd boot, login, and shutdown. I noticed the e-d-s problem shortly after I started switching users / logging out of one user then into the next without a reboot.

Deactivating "Evolution Alarm Notifier" in
System/Preferences/Sessions/Startup Programs solves the problem for me!

I tested this on several systems!

---
I vote for replacing the dependency on Evolution in ubuntu-desktop with
a dependency on a virtual "Mail Client" package! Evolution just did too
much bad stuff for too long now!

James Tait (jamestait) wrote :

Disabling Evolution Alarm Notifier in Session Startup works around the problem. Re-ordering it to priority 60 instead of 50 only delays it.

David Hajage (eusebe) wrote :

There were an update of evolution and evolution-data-server yesterday. But this morning (2008/05/06), the bug is no solved.

I just upgraded to Ubuntu 8.04 Hardy Heron. I have the evolution-data-server-2.22 process that is eating max CPU. I don't use Evolution, never started it, and don't plan to. I have been able to kill the process from the system monitor.

For you info I'm not on a dual core CPU. My laptop (Sony Vaio VGN-TX3XP) is running on an Intel Core Solo 800 Mhz (stepping to 1.2 GHz).

C de-Avillez (hggdh2) wrote :

BYPASS -- it seems to work, at least we got some positive responses, and no negative ones (Gnome users):

1. Go to System/Preferences/Sessions;
2. select the "Startup Programs" tab;
3. unselect the "Evolution Alarm Notifier" tab;
4. select the "Current Session" tab;
5. click on the "program" header; this will order the entries alphabetically;
6. highlight *all* e-a-n entries (there might be more than one), and click on the Remove button;
7. Click on the Apply button;
8. close the Sessions applet.

9. Reboot (not really needed, but just in case).
10. login again. You should be free of the E-D-S loop.

DRAWBACK: no calendar notification is available (e-a-n is the program that checks for it).

THIS IS NOT A SOLUTION, JUST A BYPASS.

Markus Kienast (elias1884) wrote :

Please disable automatic startup of "evolution alarm notifier" until this is fixed!

Some people like me are maintaining large LTSP environments and this damn thing screws up the hole system!

* When people log out, there are left over processes.
* They are unable to login again until I kill these processes.
* Alarm Notifier consumes all CPU

I attached a ps axu showing lots of user processes left over after logout. Non of the users listed was logged in, when this file was created. Pay special attention to users 1526, 1528 (names were to long to display) and haasd. Most of their processes did not stop on logout.

They can not log in anymore now! Until I get my hands on and kill their stuff.

The other users show the usual suspects:
scheidlc 17439 0.0 0.2 5340 2216 ? S 11:42 0:00 /usr/lib/libgconf2-4/gconfd-2 12
scheidlc 7864 0.0 0.2 31992 3048 ? Ssl May06 0:00 /usr/lib/bonobo-activation/bonobo-activation-server --ac-activate --ior-output-fd=16
scheidlc 7926 0.0 0.5 42448 5724 ? Sl May06 0:00 /usr/lib/evolution/evolution-data-server-2.22 --oaf-activate-iid=OAFIID:GNOME_Evolution_DataServer_CalFactory:1.2 --oaf-ior-fd=22

This bug goes hand in hand with bug #90258. Two different problems, yes. But the same source.

Disabling Evolution Alarm Notifier solved this problem on my Laptop.
Having certain calendars enabled in Evolution especially the Birthday and Weather stuff seems to trigger some misbehaviour too, especially on the logout front.

I vote for getting rid of evolution all together. But as a quick fix, disable Evolution Alarm Notifier by default!

System info:
LTSP: 1 AMD K7 CPU (32bit)
Laptop: Some 32bit Celeron CPU pre DualCore

It would be so much easier and more accurate if I could just point you to the HW profiles of my systems, which could be attached to my launchpad profile if Canonical would fund and implement an idea outlined here:

http://brainstorm.ubuntu.com/idea/1497/
https://bugs.launchpad.net/malone/+bug/3382

Bruce Edge (bruce-edge) wrote :

For me this bug happens as soon as I click the "new message" button. I don;t even have to type anything in the new msg window.

top - 09:26:56 up 2 days, 35 min, 11 users, load average: 1.40, 0.98, 0.90
Tasks: 226 total, 2 running, 224 sleeping, 0 stopped, 0 zombie
Cpu(s): 41.0%us, 21.7%sy, 0.0%ni, 37.0%id, 0.0%wa, 0.2%hi, 0.2%si, 0.0%st
Mem: 2073768k total, 1997252k used, 76516k free, 216664k buffers
Swap: 4192924k total, 406204k used, 3786720k free, 1033256k cached

  PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
11125 bedge 20 0 48216 11m 8872 S 48 0.5 0:38.84 evolution-excha
11114 bedge 20 0 128m 43m 20m S 42 2.1 0:36.04 evolution
9058 bedge 20 0 7592 4140 1456 R 32 0.2 499:16.95 gconfd-2

Note that gconfd-2 also jumps permanently onto the ready list.

To recreate: start evolution, click "new message" button.

Just like everyone else, have run --force-shutdown to get it to stop.

Here's the evolution runtime info:

11114 pts/11 Sl+ 1:17 evolution
11121 ? Sl 0:00 /usr/lib/evolution/evolution-data-server-2.22 --oaf-activate-iid=OAFIID:GNOME_Evolution_DataServer_InterfaceCheck --oaf-ior-fd=17
11125 ? Sl 1:26 /usr/lib/evolution/2.22/evolution-exchange-storage --oaf-activate-iid=OAFIID:GNOME_Evolution_Exchange_Component_Factory:2.22 --oaf-ior-fd=18
11135 ? Sl 0:00 /usr/lib/evolution/2.22/evolution-alarm-notify --oaf-activate-iid=OAFIID:GNOME_Evolution_Calendar_AlarmNotify_Factory:2.22 --oaf-ior-fd=19

0 #> uname -a
Linux ice 2.6.24-16-generic #1 SMP Thu Apr 10 13:23:42 UTC 2008 i686 GNU/Linux

Dual core Xeon:

processor : 0 (1 is the same)
vendor_id : GenuineIntel
cpu family : 6
model : 15
model name : Intel(R) Xeon(R) CPU 5160 @ 3.00GHz
stepping : 11
cpu MHz : 2992.628
cache size : 4096 KB
physical id : 0
siblings : 2
core id : 0
cpu cores : 2
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 10
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc arch_perfmon pebs bts pni monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr dca lahf_lm
bogomips : 5989.16
clflush size : 64

I disabled calendar alarm actions for all calendars. The only active plugins are :
exchange operations
local addr
local cal
mailing list act
mail to *
mark all read
save attach
select one source
subj threading

C de-Avillez (hggdh2) wrote :

@Thag: your issue does not match this bug. Please open a new bug for it (I am not aware of any current bugs dealing with evolution-exchange-storage looping).

Please have a look at http://www.gnome.org/projects/evolution/bugs.shtml, and follow the instructions on the "Exchange Connector Debug".

Thank you.

C de-Avillez (hggdh2) wrote :

@Elias Humbolt: I am confused. What does e-a-n have to do with that many processes being left running when the user logs off? Also, it is not e-a-n that consumes CPU, but e-d-s, at least on this bug. This bug deals with a loop on e-d-s on login. Any other loops, on any *other* programs, is not for this bug.

Please open a new bug for your issue; one bug per issue, since it seems you are describing more than one bug.

Finally, your vote for getting rid of evolution is of no effect here. The proper channel is the Ubuntu Development or the Ubuntu Development discussion mailing lists.

Markus Kienast (elias1884) wrote :

If you disable e-a-n the 100% cpu usage on startup described in this bug
goes away for me! So CERTAINLY e-a-n does have something to do with this
bug. Seems like you guys are looking at the wrong place.

And since disabling e-a-n does also solve the problem with the leftover
e-d-s and gconf processes described in the other bug I mentioned, it is
even more likely that e-a-n is the real sucker here! Seems to me, like
e-a-n is a piece of crap and makes all kind of other services depending
on it or services it depends on go cracy.

That is the reason I am bringing e-a-n in the play here. Because there
must be a reason the problem goes away, when I disable it.

No offence, but I think this should be enough reason to investigate in
this direction.

Elias

Dragoneye (dragoneye) wrote :

I added holidays for norway as an web ical
webcal://www.scapegoat.no/calendar/norwegian.php
to evolution->calendar
That was when my looping bug began with e-d-s.

My workaround was:

- entered property for my "Norwegian holiday ical"
   - checked "Copy calendar content locally for offline operation"
   - Refresh: 43 weeks, (previously 0 minutes, maybe thats the reason for loop?)

File->Work offline
File->Work online

got a question on synchronize witch i did. Now both my clock/alarm/calendar and all of evolution behaves as it is supposed to.

Dont know if this can lead to fixing the bug. But anyway this is my 2 cents to the pot ;-)

This bug stopped for me about 2 weeks ago.

I have proposed updates enabled and am running Hardy.

On Thu, May 15, 2008 at 9:29 PM, brettalton <email address hidden> wrote:
> This bug stopped for me about 2 weeks ago.
>
> I have proposed updates enabled and am running Hardy.
>
I can confirm that there are no more CPU usage issues after upgrading
to the proposed eds.
I rebooted my laptop at least twice a day for the past two weeks and I
haven't seen eds use much of the cpu. Whenever I looked at "ps ax|grep
evo..." it's always in "Sl" mode.

--
Lucian

David Hajage (eusebe) wrote :

The bug is not fixed for me this morning.

Tilly (henrik-digitilly) wrote :

For the record I can report that the bug has dissapeared from both my system the past week. (Linux 32, Intel/AMD dual core systems, Ubuntu 8.04).

David Hajage (eusebe) wrote :

I have activate proposed updates.

After reboot, the bug is NOT fixed on my laptop (Dell xps M1210).

David Hajage (eusebe) wrote :

The bug is still not fixed.
I noted that the bug appeared also when I click on the calendar of my panel, but only when my laptop is offline.

Thus, I uninstalled evolution-data-server and its dependancies : evolution and ekiga (?). I liked to see the events in my calendar on my desktop, and I'm very sad to see this functionality broken in hardy LTS.

laalamz (tom-jerry-hacker) wrote :

M using linux64, intel, 1cpu

100% usage is taking place after i have done the proposed updates

****not fixed****

Grizzly (sven-witterstein) wrote :

Not much activity here on this bug report any more. I think some recent updates have changed the bug or obscured the supposed race condition. Amilo A1667G with Turion x64 does not have the issue any more, Athlon64x2 does not have it, and new Phenom x4 does not have it. Not possible to reproduce it here, online or offline while booting/logging in or later (that is routers on/off power, wlan bridge to living room with or without power (dhcp) - doesnt make a difference).
Maybe on single cores more noticable? I am off this bug, as I cant reproduce it.

David Hajage (eusebe) wrote :

I have uninstall evolution until today, just to see if this bug was fixed.

On my Dell XPS M1210 (Intel(R) Core(TM)2 CPU T5500 @ 1.66GHz), the bug is not fixed.

Adam Niedling (krychek) wrote :

pépére: mentioning the version of Ubuntu you're using would be more useful than the frequency of you processor.

Christopher Berner (cberner) wrote :

I have Ubuntu 8.10 x64, an Intel E6400, and I'm having this problem.

driedfruit (driedfruit) wrote :

linux32, intel, 2cpu

VexaAE (vexaae) wrote :

Intrepid x32 on Intel T7100 (dual core laptop).

On my machine, it doesn't go to 100% when i login. It usually hogs CPU after using Evolution to send emails (i haven't found it hogging CPU when reading emails). Also, it only hogs AFTER I press send.

If i'm impatient I kill evolution-data-server when it goes berserk, but if you don't kill it and wait a couple minutes, it actually goes back to normal operation. It is quite a nuissance though.

jaunty amd64 on intel core duo on a Dell Vostro 1510 laptop,

same : the process goes to 100% when I want to read the first mail, if I don't wait a long time this mail is still "formatting". When the first one is read, others sometimes goes well...

{linux32, Intel, 1 CPU}

IBM Thinkpad T42p Pentium M 1.8Ghz/2Mb
Ubuntu 9.04 Jaunty Jackelope alpha
2.6.28-8-generic
evolution 2.25.91

Since this is on a laptop I have quite a few different network setups, i.e. several WLANs, normal Ethernet, UMTS via usb. If any additional data on this could be helpful, let me know. I'll try to see, whether I can find a relation between this and which network I use.

This time, I noticed the problem after a wake-up from sleep status.

Nick Twigg (nick-nick-web) wrote :

Linux64, AMD64, 2CPU's confirmed.

Ubuntu Jaunty 9.04 b5

{linux32, Intel, 2CPU}

IBM Thinkpad T60 Intel Core2Duo 1.8Ghz / 3Mb
Ubuntu 8.04 (LTS)

It's not allways using 100% of both CPU but 100% of one in any case.

The problem ist constant (by every bootup).
I have to "interupt the process" to be able to use the computer.

No problem on my Desktop {linux64, Asus, 2CPU}

Changed in evolution-data-server:
status: Confirmed → In Progress

Hi,

I'm using a thinkpad T60 {linux32, Intel core2duo, 2CPUs}. I have no
problem anymore since upgrading from Ubuntu Hardy Heron to Jaunty
Jackalope...

I hope it's helping !

Graham F (gandalfmeister) wrote :

Karic Release candidate - evolution data server maxed out at 100% and memory usage skyrocketed - at 1.0Gb and still climbing at the moment. I have 1 gmail calender that I tied to look at., and 2 gmail IMAP accounts. Eeek.

sam tygier (samtygier) wrote :

is anyone still seeing this? i have not for a while, and the upstream bug is marked fixed.

Changed in evolution-data-server (Ubuntu):
status: Triaged → Incomplete
James Tait (jamestait) wrote :

If it is still happening, it hasn't brought my machine to a standstill as it used to do. That said, I've had two re-installs since then, so that changes the picture for me anyway.

Changed in evolution-data-server:
importance: Unknown → Critical
status: In Progress → Expired
James Lewis (james-fsck) wrote :

I have not seen this happen in 10.04 or 10.10..

I have not seen ths happen in 10.04 (32 and 64 bit) too !

2010/9/16 James Lewis <email address hidden>

> I have not seen this happen in 10.04 or 10.10..
>
> --
> [MASTER] E-D-S hangs on login and uses 100% cpu
> https://bugs.launchpad.net/bugs/151536
> You received this bug notification because you are a direct subscriber
> of the bug.
>
> Status in Evolution Data Server: Expired
> Status in “evolution-data-server” package in Ubuntu: Incomplete
>
> Bug description:
> Binary package hint: evolution-data-server
>
> <PLEASE> -- do not add simple "me too" comments, they do not help: we
> already *know* this is a problem, and these comments only make it more
> difficult to find potentially important data in the bug. If you have the
> knowledge and time to help debug, your help *is* appreciated. Go to
> #ubuntu-bugs in irc.freenode.net, and discuss it there. I am sure that any
> one of us will be more than happy to get on with this ;-)</PLEASE>
>
> On login evolution-data-server uses 100% cpu. At login
> evolution-alarm-notify is auto-started by gnome-session; e-a-n kicks in
> E-D-S and, for yet an unknown reason, E-D-S enters a loop. strace on E-D-S
> shows a sequence of very fast poll() calls; a strace on e-a-n shows e-a-n
> waiting for an event. This seems to happen on some machines, but not all. It
> affects both Evolution users and non-users -- E-D-S is still installed under
> Gnome even if you do not use the Evolution mail client.
>
> All symptoms are of a race condition, very early in E-D-S or a library
> startup. Backtraces are inconclusive, and show slightly different code
> paths. I cannot reproduce it on my machine (an AMD64 dual-core), no matter
> what I try.
>
> I would like to know what Linux and hardware architecture, and how many
> CPUs for those that are *frequently* hit (for example, {linux64, AMD64, 2
> CPUs}, meaning I am running linux64 on an AMD64 with two CPUs; {linux32,
> AMD64, 4 CPUs}, meaning I am running linux32 Intel on an AMD64 with 4
> CPUs).
>
> Right now we are considering this as a race very early on session startup,
> perhaps caused by the current placement of e-a-n on the default session list
> (with the default priority of 50). I myself still cannot reproduce it, and I
> wonder if having a dual-CPU machine, and/or linux64, and/or AMD64, makes the
> race more difficult to happen.
>
> To unsubscribe from this bug, go to:
> https://bugs.launchpad.net/evolution-data-server/+bug/151536/+subscribe
>

--
NATUCOM
Gérald Fringeli
Avenue de la Gare 41
CH - 2800 Delémont

Tel : +41(0)44.586.12.19
Fax : +41(0)32.423.12.17

<email address hidden>
www.natucom.ch

This but is affecting me on a dual core (with a lot of empty hard disk space) hp pavillion dv9000, Ubuntu 10.04.
uname -a
Linux Balenotterax 2.6.32-25-generic #45-Ubuntu SMP Sat Oct 16 19:52:42 UTC 2010 x86_64 GNU/Linux

Thank you,
Ricky

Evan Peck (colors) on 2012-01-17
Changed in evolution-data-server (Ubuntu):
status: Incomplete → Confirmed
Sebastien Bacher (seb128) wrote :

this bug is old and didn't get duplicates in years, upstream closed it as obsolete, doing the same, if you still get issues better to register a new bug report

Changed in evolution-data-server (Ubuntu):
assignee: Ubuntu Desktop Bugs (desktop-bugs) → nobody
status: Confirmed → Invalid
Displaying first 40 and last 40 comments. View all 161 comments or add a comment.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.