Bug 666603 - openoffice.org does not start - includes solution to issue
Summary: openoffice.org does not start - includes solution to issue
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Fedora
Classification: Fedora
Component: openoffice.org
Version: 14
Hardware: x86_64
OS: Linux
low
low
Target Milestone: ---
Assignee: Caolan McNamara
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2011-01-01 11:55 UTC by Harish Pillay
Modified: 2011-03-16 16:13 UTC (History)
5 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2011-03-16 16:13:05 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)

Description Harish Pillay 2011-01-01 11:55:12 UTC
Description of problem:
When trying to double click on a .odp file, openoffice.org starts up but fails before fully activating.  Does not even trigger a ABRT incident.

Version-Release number of selected component (if applicable):
1. openoffice.org-core-3.3.0-18.2.fc14.x86_64
2. Linux vostro.sin.redhat.com 2.6.35.10-74.fc14.x86_64 #1 SMP Thu Dec 23 16:04:50 UTC 2010 x86_64 x86_64 x86_64 GNU/Linux

How reproducible:
1. Double click on a .odp file
2. System shows that ooimpress is being started, but nothing happens and ooimpress fails silently.

Steps to Reproduce:
1. Double click on a .odp file or from a terminal, run: ooimpress file.odp
2. there will be some activities and then it fails silently
  
Expected results:
1. ooimpress should have started and offered the file for edit.

Additional info:
1. while logged in as a normal user, within the desktop, start a terminal.
2  in the terminal, su - to root
3. start ooimpress file.odp
4. ooimpress starts correctly and all is well.
5. this suggested that openoffice.org is OK, just that there is some issue
   within the user that I was logged in.
6. checking /tmp, I found a file:
      OSL_PIPE_500_SingleOfficeIPC_9b6fe25a74ede9679728154c992740a1
7. rm'ed the file and then as a normal user, ooimpress file.odp started normally.
8. Looks like for some reason, the residual file in /tmp, OSL* above, which was created two days ago was causing the issue. Removing it solved the problem.

Comment 1 Orion Poplawski 2011-01-05 16:42:31 UTC
I just ran into the same thing.  Running ooffice from the command line did trigger an abrt crash report.  May be the cause of bug 667082 and bug 667404.

Comment 2 Harish Pillay 2011-01-06 14:28:26 UTC
(In reply to comment #1)
> I just ran into the same thing.  Running ooffice from the command line did
> trigger an abrt crash report.  May be the cause of bug 667082 and bug 667404.

Did you try to start ooffice as root or another user?

Harish

Comment 3 Orion Poplawski 2011-01-06 15:14:08 UTC
The problem only affected one user - another user could run ooffice just fine.  Never ran it as root.  Removing the /tmp IPC pipe fixed it for me as well.

Comment 4 Harish Pillay 2011-01-06 15:20:30 UTC
(In reply to comment #3)
> The problem only affected one user - another user could run ooffice just fine. 
> Never ran it as root.  Removing the /tmp IPC pipe fixed it for me as well.

Good to know that the solution seems to be removal of that IPC pipe in /tmp.

I am testing out LibreOffice as well and have not seen this issue yet.

Comment 5 Stephen P. Schaefer 2011-01-11 20:18:21 UTC
I experienced this as well.  I've annotated some lines with # comments

bash-4.1$ oocalc
# nothing happened
bash-4.1$ echo $?
0
bash-4.1$ ls /tmp/O*
/tmp/OSL_PIPE_4121_SingleOfficeIPC_a7279bf8114e9fba64b64ad0f3c26cd6
bash-4.1$ ls -l /tmp/OSL*
srwxrwxr-x. 1 sps sps 0 Jan  6 23:24 /tmp/OSL_PIPE_4121_SingleOfficeIPC_a7279bf8114e9fba64b64ad0f3c26cd6
bash-4.1$ oowriter
# nothing happened
bash-4.1$ echo $?
0
bash-4.1$ ps auxwww | grep oowriter
sps      26551  0.0  0.0 103416   840 pts/3    S+   15:11   0:00 grep oowriter
bash-4.1$ rm /tmp/OSL*
bash-4.1$ oowriter
# oowriter starts just fine
bash-4.1$ ls -l /tmp/OSL*
ls: cannot access /tmp/OSL*: No such file or directory
# ...and cleans up after exiting normally

I suspect that some forms of openoffice exit (e.g., sudden power loss) leave behind the pipe file.  Possible workaround: on openoffice component startup, check that there is a process listening to the pipe (with, e.g., lsof) and remove it if not.

Comment 6 David Tardon 2011-01-12 07:09:15 UTC
> I suspect that some forms of openoffice exit (e.g., sudden power loss) leave
> behind the pipe file.
That's right, but it's not a problem.

> Possible workaround: on openoffice component startup,
> check that there is a process listening to the pipe (with, e.g., lsof) and
> remove it if not.
But that's not what you did. Otherwise you would probably have found that
there indeed was a process listening on the socket...

(In reply to comment #5)
> I experienced this as well.  I've annotated some lines with # comments
> 
> bash-4.1$ oocalc
> # nothing happened
> bash-4.1$ echo $?
> 0
> bash-4.1$ ls /tmp/O*
> /tmp/OSL_PIPE_4121_SingleOfficeIPC_a7279bf8114e9fba64b64ad0f3c26cd6
> bash-4.1$ ls -l /tmp/OSL*
> srwxrwxr-x. 1 sps sps 0 Jan  6 23:24
> /tmp/OSL_PIPE_4121_SingleOfficeIPC_a7279bf8114e9fba64b64ad0f3c26cd6
> bash-4.1$ oowriter
> # nothing happened
> bash-4.1$ echo $?
> 0
> bash-4.1$ ps auxwww | grep oowriter
> sps      26551  0.0  0.0 103416   840 pts/3    S+   15:11   0:00 grep oowriter
That would be pretty conclusive if the actual binary were named oowriter. Unfortunately, it's not. What's the output of either
lsof /tmp/OSL_PIPE_4121_SingleOfficeIPC_a7279bf8114e9fba64b64ad0f3c26cd6
or
ps -C swriter.bin
?

Comment 7 Stephen P. Schaefer 2011-01-12 18:46:48 UTC
Remembering that things are now working correctly, this is what I get:

bash-4.1$ lsof /tmp/OSL_PIPE_4121_SingleOfficeIPC_a7279bf8114e9fba64b64ad0f3c26cd6
COMMAND    PID USER   FD   TYPE             DEVICE SIZE/OFF  NODE NAME
soffice.b 3148  sps    8u  unix 0xffff88009254c000      0t0 22889 /tmp/OSL_PIPE_4121_SingleOfficeIPC_a7279bf8114e9fba64b64ad0f3c26cd6

Comment 8 Orion Poplawski 2011-01-14 16:37:38 UTC
In our case, I wonder if having multiple sessions (desktop + vncserver) on a single machine is causing this.  Can OOo handle this?

Comment 9 Caolan McNamara 2011-03-16 16:13:05 UTC
If the same user is logged in multiple times then that would explain this i.e. another OOo belonging to that user running on a different display. Otherwise I can't reproduce this problem.

rm the pipe just cuts the connection so that two instances can be run, instead of the usual one, with all the consequent possible problems of munging the personal config dir as I bet there's no support to make that safe.

Assuming I'm right, then this is "just the way it work" unfortunately. If I'm wrong, i.e. someone has a reproducible scenario feel free to fill in those details and reopen this.


Note You need to log in before you can comment on or make changes to this bug.