Bug 217031

Summary: beagle sometimes consumes 100% CPU
Product: [Fedora] Fedora Reporter: Need Real Name <mal>
Component: beagleAssignee: Alexander Larsson <alexl>
Status: CLOSED RAWHIDE QA Contact:
Severity: medium Docs Contact:
Priority: medium    
Version: 6CC: alex, averma, gnomeuser, joeshaw, k.georgiou, ncunning, tadej.j, valent.turkovic, wwoods
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2007-10-21 04:27:35 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
Patch against 0.2.16.x to fix the empty OOo file
none
screenshot of beagle on Fedora rawhide desktop running ok. none

Description Need Real Name 2006-11-23 09:07:16 UTC
beagle-update-in 
beagle-helper
sometimes consume 100% CPU, making computer 
slow responding
rpm -q beagle
beagle-0.2.10-7.fc6

See also
http://ubuntuforums.org/showthread.php?t=281491
and other forums

Comment 1 Need Real Name 2006-11-23 09:08:11 UTC
rpm -e beagle beagle-gui beagle-evolution
fixes the problem

Comment 2 Need Real Name 2006-11-23 09:15:43 UTC
After beagle removal the FC6 feels so much faster,
I thought FC6 was slow compared to FC5 because of new gnome bloat,
but beagle removal totally changed the impression.
It feels now like 3 times CPU increase.
And no mysterious stalls any more.

I would strongly recommend to remove beagle altogether.
It's limited functionality does not worth 
ruining FC6 overall impression.


Comment 3 Need Real Name 2006-11-23 09:19:38 UTC
There is also huge win in X startup.
Now the command 
startx
brings the X desktop at least five times faster.

Comment 4 Alex Lancaster 2006-11-28 04:17:45 UTC
I get the same problem, 100% CPU usage.  It appears that beagled-helper somehow
hangs because I need to:

killall -9 beagled-helper
killall -9 beagled

to stop it.  This is not bug #212370 because it doesn't "hang" in the same way
and I've already upgraded to beagle-0.2.10-7.fc6.

I am using a suspend2  kernel from http://mhensler.de/  if that makes a
difference.  I have noticed that sometimes starts after I've resumed.  But
there's no real pattern to be able to duplicate it reliably.

With all due respect to comment #1 and comment #2 removing beagle isn't a
solution, it's a temporary workaround at best.

Comment 5 Alex Lancaster 2006-11-28 04:19:04 UTC
This is actually probably a duplicate of bug #205258.

Comment 6 Alexander Larsson 2006-11-28 13:22:04 UTC
Please test the beagle-0.2.13-1.fc6 update in updates-testing. Maybe it helps.


Comment 7 Michael De La Rue 2007-01-14 12:04:29 UTC
According to the Beagle FAQ http://beagle-project.org/Troubleshooting_CPU

"Beagle version 0.2.13 and older are buggy"

"The first thing is to ensure that you're running at least Beagle 0.2.14 or
newer. [...] Bugs in older versions existed that could cause directories to be
continuously reindexed."

Probably worth rebuilding on the basis of a newer Beagle?

Comment 8 David Nielsen 2007-05-10 01:31:54 UTC
Still happens on FC7test4 x86_64 using beagle-0.2.16.2-4.fc7, it spins one CPU
constantly. I'm on a dual-core machine but it only spins the one CPU sp I didn't
notice it till now.. wow.

Adding FC7Blocker

Comment 9 David Nielsen 2007-05-10 01:38:30 UTC
Following the guide at http://beagle-project.org/Troubleshooting_CPU we find that:

20070510 03:07:53.5609 06808 IndexH ERROR: Unable to open
/home/dnielsen/Documents/New Document.odt.  Probably an invalid OpenOffice document.

is causing this, I believe this bug is fixed in 0.2.17 according to the changelog:

http://mail.gnome.org/archives/dashboard-hackers/2007-May/msg00013.html

* Fix an infinite loop problem with empty OpenOffice documents. [bgo
 #427256, #435694]

This also contains memory use and performance enhancements so consider this a
request for an update from your friends on the QA team.

Comment 10 Matthias Clasen 2007-05-11 14:43:38 UTC
Should probably be handled as a post f7-update at this point

Comment 11 David Nielsen 2007-05-11 15:20:53 UTC
0.2.17 also contains a 100% cpu load bug on my machine, Beagle it seems is just
conspiring to make our default install suck. Would it be sane to disable it in
our default install, I know it late in the game but shipping with a 100% load
bug is bad and there's no fix we can deploy post release as of yet for that what
one. I don't think relying on post release updates here is safe.

Comment 12 Need Real Name 2007-05-11 19:03:28 UTC
I would suggest just to remove beagle.
The FC6 was just unusable with beagle running.
At least do not ruin FC7 impression.
I would repeat it again about beagle:

It's limited functionality does not worth 
ruining FC overall impression.

Comment 13 Rahul Sundaram 2007-05-13 18:39:27 UTC
(adding wwoods)

Either request a exception from rel eng to fix this bug by providing an update
before the general release or we should just disable it by default for now and
have this in the "known issues" section of the announcement and/or release notes. 

Can we have a final decision on this soon?

Comment 14 Alexander Larsson 2007-05-14 14:30:19 UTC
I really don't have a lot of time to work on beagle, and it seems to keep having
these kinds of bugs. We should probably disable it. I think not installing it by
default might be the right thing to do.


Comment 15 Alexander Larsson 2007-05-14 15:05:02 UTC
I removed beagle from the default install in fedora 7, as bugs like this makes
the distro look pretty bad, and they keep turning up.

Comment 16 Valent Turkovic 2007-05-16 09:02:26 UTC
I use beagle on Fedora Core 6 and Fedora 7 test 4 system all the time and have
ZERO problems.

I had problems on Fedora 5 while testing beta versions of beagle. After FC6
launch I had only great experience with it...

Comment 17 Joe Shaw 2007-05-18 21:25:49 UTC
Created attachment 155029 [details]
Patch against 0.2.16.x to fix the empty OOo file

If you don't want to update to 0.2.17, I've attached a patch which fixes the
empty OOo file problem.  We apply this to 0.2.16.3 in SUSE.

Comment 18 Joe Shaw 2007-05-18 21:28:27 UTC
(In reply to comment #11)
> 0.2.17 also contains a 100% cpu load bug on my machine

I can work with you on this one, but I didn't know about it until I happened to
just come across it now.  What file is causing the issue?


Comment 19 David Nielsen 2007-05-18 22:57:47 UTC
I attempted to follow the troubleshooting guide but the process that runs
rampant is not the helper one but beagled itself. There's no indication of error
in the log but if one looks at what beagle is doing using the status program it
appears that it is indexing mail. I think the sanest approach is to wait for the
package to officially be updated once Development opens again and then have a
look at matters.

Seeing as Beagle is frozen for F7 I'd rather encourage the testers to work with
you on making Beagle rock in F8, if you have the excess energy comaintaining the
Beagle package would be good in that it would allow you to push patched
versions. Having Beagle be useful during as much of the development cycle
hopefully give you better feedback from the Fedora userbase so to increase rockage. 

Comment 20 Joe Shaw 2007-05-21 13:57:35 UTC
Interesting, I haven't seen beagled itself take up 100% CPU in a long, long time.

The one reason I can think of that it might is the screensaver case.  Checking
your logs (~/.beagle/Log/current-Beagle) would help too, as you might be hitting
an exception unexpectedly?

Comment 21 Valent Turkovic 2007-05-21 18:38:33 UTC
Joe and David if you need some to test beagle with Fedora 6 and 7 I'm
volunteering for testing new versions - just give me instructions on how I need
to test it and in which way to give you feedback. 

Comment 22 Valent Turkovic 2007-05-30 12:28:11 UTC
I can share my experience with beagle from over several months:

ibm R52 laptop + fedora 6 + beagle + deskbar = no cpu issues
HP nx7300 laptop + fedora 6 + beagle + deskbar = no cpu issues
HP nx7300 laptop + fedora 7 + beagle + deskbar = no cpu issues (few weeks only)
noname desktop + opensuse 10.2 + beagle + deskbar = no cpu issues
HP laptop of my girlfriend + linux mint (ubuntu edgy) + beagle = no cpu issues
HP laptop of my girlfriend + linux mint (ubuntu feisty) + beagle = no cpu issues
(few weeks only)

Comment 23 Valent Turkovic 2007-07-03 12:55:00 UTC
Hi I would help with beagle project - I would like to help out and
make it installed by default in Fedora 8.

I have a few Fedora 7 installs and one Fedora Core 6, so if you send
me all problematic files with which ones you had problems I can test
it and send you feedback.

Can you tell me what is the status od beagle now? Are these bugs fixed
by now and will beagle be installed by default in Fedora 8?

Comment 24 David Nielsen 2007-07-03 13:07:10 UTC
This bug has been fixed but at least in Rawhide Beagle is being problematic.. again:

See https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=245521

Comment 25 Valent Turkovic 2007-07-03 14:15:30 UTC
Correct me if I'm wrong. I see that rawhide beagle bug is reported only for
64bit platform. Does the 32bit version work ok?

Comment 26 Valent Turkovic 2007-07-03 21:16:13 UTC
Created attachment 158476 [details]
screenshot of beagle on Fedora rawhide desktop running ok.

screenshot of beagle on Fedora rawhide desktop running ok.

Comment 27 Valent Turkovic 2007-07-03 21:18:02 UTC
I have updated Fedora test 4 to latest rawhide and beagle works just fine. Check
out the screenshot. I do get an strange output from beagle-info but it works
just fine.

$ beagle-info --daemon-version

***MEMORY-WARNING***: /usr/lib/beagle/Info.exe[3290]: GSlice: g_thread_init()
must be called before all other GLib functions; memory corruption due to late
invocation of g_thread_init() has been detected; this program is likely to
crash, leak or unexpectedly abort soon...
Daemon version: 0.2.17


Comment 28 Joe Shaw 2007-07-03 21:37:08 UTC
(In reply to comment #27)
> ***MEMORY-WARNING***: /usr/lib/beagle/Info.exe[3290]: GSlice: g_thread_init()
> must be called before all other GLib functions; memory corruption due to late
> invocation of g_thread_init() has been detected; this program is likely to
> crash, leak or unexpectedly abort soon...

I believe this is a problem with gtk#, but I will have to investigate further. 
In any case, it's unrelated to this bug (100% CPU utilization).

Comment 29 Valent Turkovic 2007-07-04 06:21:45 UTC
It probably is a gtk bug, but I posted that only for you to see the version of
my beagle. I belive my post with attachement is vital to this but in the way it
shows that on 32bit fedora rawhide system beagle works without 100% CPU bug - as
you all can see.

Comment 30 Joe Shaw 2007-07-04 14:04:45 UTC
(In reply to comment #29)
> I belive my post with attachement is vital to this but in the way it
> shows that on 32bit fedora rawhide system beagle works without 100% CPU bug - as
> you all can see.

Well, part of the problem is that certain situations or data are evidently
causing the CPU issue.  See above about the empty OOo file for an example.

The title of the bug is "beagle sometimes consumes 100% CPU", not "beagle always
consumes 100% CPU". :)

As always, the info on the wiki http://beagle-project.org/Troubleshooting_CPU is
helpful in debugging these issues.

I've also been working on a 0.3.0 version which will more strictly require
inotify (and warn if unable to get enough watches) so that the problem of
looping over directories w/o inotify watches will go away.  That's a common
cause of high CPU utilization when the screensaver is on.

Comment 31 Valent Turkovic 2007-07-06 08:44:54 UTC
Great! I can't wait to test it out. Will it be also available for FC6 as for F7?

Comment 32 David Nielsen 2007-10-21 04:27:35 UTC
I've been running Beagle on F8 for a while now without hitting the 100% CPU
usage bugs and generally Beagle is very well behaved. I think the sensible thing
is to close this bug.

For F9 I plan to put Beagle under the loving embrace of the GNOME QA beat, I'd
like people to join me in testing. The end result would hopefully be having
Beagle and other Mono apps moved under the supervision of some kind of Mono SIG
so to unload Alex and ensure quicker response to bugs and version upgrades. I'll
ping Paul Johnson about this as he seems very dependant on Mono and might be
conned into maintaining Mono on Fedora.

Regardless I'm closing this bug as RAWHIDE with encouragement to upgrade to F8
as soon as it's out for an improved bugfree Beagle experience.

Comment 33 Valent Turkovic 2007-10-21 11:33:49 UTC
Great news! I'll be happy to test new beagle on F8 (I'm using it now on rawhide)
and on F9 devel when it starts. I would love to see it running in gnome under F9
so if you need bug testers just drop me a line.