Bug 103001

Summary: GConf1 needs to be updated to use local locks
Product: [Fedora] Fedora Reporter: Rob Hughes <rob>
Component: GConfAssignee: Ray Strode [halfline] <rstrode>
Status: CLOSED WONTFIX QA Contact:
Severity: medium Docs Contact:
Priority: medium    
Version: rawhideKeywords: Triaged
Target Milestone: ---   
Target Release: ---   
Hardware: i386   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2008-04-23 05:34:59 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---
Bug Depends On:    
Bug Blocks: 100643    

Description Rob Hughes 2003-08-25 02:08:47 UTC
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20030225

Description of problem:
After an update to GConf2 and -devel from rawhide, I'm unable to start any GConf
dependent app. Running oaf-slay and bonobo-slay, then cleaning up the leftover
locks worked once, but since then, I'm unable to start, for example, evolution,
grcm, etc. I attempted to run some of these from a shell directly to get some
error output, but other than the app hanging, there isn't any. I've also tried
using the GCONF_LOCAL_LOCKS trick, as well as running various programs as a
local user. The result is the same each time. I'm running RH9 + various updates
from rawhide.

Version-Release number of selected component (if applicable):
GConf2-2.3.3-2

How reproducible:
Always

Steps to Reproduce:
1. Update to GConf2-2.3.3-2 from rawhide
2. Try to start evolution
3. Observe hangs and stale lock files after killing app.
    

Actual Results:  The app fails to start

Expected Results:  The app starts.

Additional info:

When I first started, I got an error indicating an error regarding a socket was
seen, but I've been unable to reproduce this and failed to record it. I do,
however, see a number of gconfd-2 processes left over after anything is run that
uses gcond2/CORBA.

Comment 1 Rob Hughes 2003-08-25 03:37:32 UTC
I went digging through my logs, and found the error I referred to earlier:

gconfd (rob-5971): Failed to send buffer

Looks like it's having a problem with sockets or something.

Comment 2 Alexander Larsson 2003-08-27 10:08:39 UTC
Did you also update ORBit2 and bonobo?

Comment 3 Rob Hughes 2003-08-27 11:31:05 UTC
Here are my versions:
bonobo-1.0.22-7
bonobo-devel-1.0.22-7
libbonobo-devel-2.3.6-2
libbonoboui-2.3.6-2
libbonoboui-devel-2.3.6-2
bonobo-conf-0.16-5
bonobo-conf-devel-0.16-5
libbonobo-2.3.6-2
ORBit-0.5.17-10.3
ORBit-devel-0.5.17-10.3
ORBit2-2.7.5-5
ORBit2-devel-2.7.5-5

I believe these are all the current versions from rawhide.


Comment 4 Rob Hughes 2003-09-09 03:28:10 UTC
I went back, ripped out evolution, gconf, and all the bonobo stuff, removed
every trace of anything that looked like a pipe and reinstalled. Other than not
being able to exit evolution (I have to do a killev) and links failing to launch
the associated app, everything seems to be working.

Comment 5 Alexander Larsson 2003-09-10 13:01:06 UTC
Very strange. Its possible that GConf had problems with earlier versions of
ORBit, but I have no idea what caused persistant problems like this.

Comment 6 Havoc Pennington 2003-10-04 05:45:45 UTC
doesn't allow gconf _1_ apps to start, no?

gconf1 needs to default to local locks now...

Comment 7 Rob Hughes 2003-10-04 12:12:41 UTC
I had updated both gconf1 and gconf2. The list of apps included evolution and
grcm. Are those both gconf1 apps? I though evolution used gconf2.

Comment 8 Havoc Pennington 2003-10-04 16:41:27 UTC
Depends on the version of evolution, 1.2 uses gconf1, 1.4 gconf2.

I know gconf 1 apps are broken, anyhow. If gconf 2 apps are broken it's some odd
orbit thing probably, gconf 1 is more complicated.

Comment 9 Mark McLoughlin 2004-07-28 06:37:26 UTC
Okay, status seems to be:

  GConf2 has been changed to keep its locks in /tmp rather than
  in $HOME. GConf1 hasn't been updated to use local locks so
  now if you have gconfd-2 running and you do

    $> gconftool-1 -R /tmp

  then gconfd-1 gets activated rather. Broken.

Comment 10 Matthias Clasen 2008-03-26 16:10:52 UTC
I don't think GConf 1 bugs are worth keeping open at this point.