Don't know how to make this short. Please bear with me. The length
reflects that I've done a TON of work on this issue before bring it
here; if you can handle a full read of this post, we can avoid a lot of
suggestions for things I've already done, probably more than one. :)
Mdk 2006 / FF 1.0.6 (I still had it installed), FF 1.5.x (I'd been using
that), FF 2.0.0.3 now installed (others removed; see below)
Yesterday FF started crashing (disappearing), first displaying for a
just an extremely brief flash of the title bar (only) of what I'm going
to call a dialog box. I can repeat this with 100% predictability.
It happened with both an old copy I still had of 1.0.6 and, worse, with
the 1.5.x I'd been using.
But here are the kickers. This only happens for user blinky. Another
user I have set up (forever, not just as a test for this issue) can run
FF fine.
Here's what I've done.
I've removed every moz and FF related file from blinky's home directory.
I've removed - by hand - files from all previous versions of FF in
non-home directories: /usr/[whatever] and even docs in share. I spent
most of last night removing FF stuff.
I installed 2.0.0.3 (whatever was current yesterday) from a tarball.
Clean installation. New. Shiny. Still had the red bow on it.
Problem persisted.
If I remove /home/blinky/.mozilla, FF crashes as described on trying to
run it.
If I copy /home/[other user]/.mozilla to /home/blinky, I can run FF
until something pops up a dialog box (or whatever you'd call an
internal pop-up that was not a tab or a separate FF window). My quick
test is Help/About -- that kills FF every time. And once this has
happened (using the .mozilla dir from another user), FF dies on startup
-- until I again nuke /home/blinky/.mozilla and re-copy over the one
from the other user; i.e., I get one run, and one crash, out of that
imported .mozilla dir.
I've been through X restarts, sesssion closes and reopens, and cold
reboots.
What the hell can be sticking around doing this?
The one idea I can throw out is something I did with an extension before
all this started happening (but I don't remember if it was *directly*
before -- was just an extension I'd had for a couple days that I'd been
tinkering with. I had installed the Firefox Showcase extension, which
creates a tab or a "window" (not a new full-FF window, but a window
within the FF instance that's running, like a pop-up or a dialog box,m
and puts in it thumbnails of the web pages you have tabbed. Quite nice,
and very configurable. But one of the settings - *possibly* the last one
I tried - was, basically, "use a window, not a tab, and keep that window
open even if no browsers are running". I know my last setting for that
extension was to use a "window", not a tab. I remember "window shading"
that window (it has no minimize and it was in the way). Now, remember
me saying that when FF dies, it flashes a title bar for a few
milliseconds before disappearing? Well, that might be the "window
shaded" Showcase window.
BUT, as I said, I removed everything FF-related and installed a clean
new version. How the hell could something like that persist? I've
certainly not got that extension reloaded -- I couldn't if I tried,
becauese the "allow this to be installed?" box crashes FF. :)
Here's an error from opening FF with OtherUser's .mozilla dir and then
clicking Help/About:
The program 'Gecko' received an X Window System error.
This probably reflects a bug in the program.
The error was 'BadAlloc (insufficient resources for operation)'.
(Details: serial 6379 error_code 11 request_code 53 minor_code 0)
(Note to programmers: normally, X errors are reported asynchronously;
that is, you will receive the error a while after causing it.
To debug your program, run it with the --sync command line
option to change this behavior. You can then get a meaningful
backtrace from your debugger if you break on the gdk_x_error() function.)
You *really* realize how good FF is when all can't use it. :)
--
Blinky RLU 297263
Killing all posts from Google Groups
The Usenet Improvement Project: http://blinkynet.net/comp/uip5.html
Right there, that tells you that your system installation works.
That assumes you do not have a local firefox install in that "other account"
No need to be screwing around with the system wide installed files.
It has to be something local to blinkey's account.
Go back and verify firefox still works before trying to debug blinky's
problem. Check Help About Mozilla to verify which firefox is running
in that "other account".
The thunderbird/firefox applications directories involved in /home/$USER are
.fullcircle, .thunderbird, .mozilla and .firefox
Now if you did a
/bin/rm -fr .fullcircle .mozilla .firefox
ls -al .fullcircle .mozilla .firefox
shows no files found, and next firefox launch still crashes,
I would then assume your desktop manager has something causing your problem.
That assumes you can create those directories in blinky's directory
and they are owned by blinky's UID/GID.
Click up a terminal, and check with the commands:
firefox
ls -al .fullcircle .mozilla .firefox
If it still crashes, you might try this 5 command long command
grep -i -r firefox .???* \
| grep -i -v .fullcircle \
| grep -i -v .mozilla \
| grep -i -v .firefox \
| grep -i -v .bash_history
If nothing shows up, except for socket error messages, then I will have
to assume you have set some global value in your desktop manager which
firefox does not like.
In that case I would logout of the desktop manager, and move the
desktop manager directory (ies) and launch the desktop manager again.
KDE Example:
cd ~blinky
pwd
mv .kde .kde_broke
Next login/launch of kde, under blinky, should create .kde and there you
should be able to click up a terminal and run
firefox
Hi, Bit. I'll comment as we go along here. Thanks for the reply. I
*really* appreciate it -- I know OP was long and tedious. I was just
trying to head off starting at the level of "Is your computer plugged
in?" ;)
> Right there, that tells you that your system installation works.
> That assumes you do not have a local firefox install in that "other account"
Right. It's in /bin/local .
> No need to be screwing around with the system wide installed files.
> It has to be something local to blinkey's account.
Until I cleared everything out I didn't know that; I didn't think to
look at another user before the great housecleaning. :)
> Go back and verify firefox still works before trying to debug blinky's
> problem. Check Help About Mozilla to verify which firefox is running
> in that "other account".
It still works for the other user, as noted in my OP. That has not
changed since posting.
2.0.0.3. The only one now installed. I thought I noted that in OP.
> The thunderbird/firefox applications directories involved in /home/$USER are
> .fullcircle, .thunderbird, .mozilla and .firefox
I only have .mozilla (which contains firefox) and .fullcircle.
Same as otheruser.
> Now if you did a
> /bin/rm -fr .fullcircle .mozilla .firefox
> ls -al .fullcircle .mozilla .firefox
> shows no files found, and next firefox launch still crashes,
> I would then assume your desktop manager has something causing your problem.
There is no .firefox.
FF crashes on opening if .mozilla is deleted as stated in OP. It runs
if otheruser's .mozilla is copied into Blinky's home dir -- but only
until it encounters one of those death-dealing title bars.
I can remove .fullcircle. That's safe?
> That assumes you can create those directories in blinky's directory
> and they are owned by blinky's UID/GID.
.fullcircle has, as far as I know, always been there I've not done
anything with it. I only noticed it when you started talking about it,
since "fullcircle" does not exactly imply "firefox" (to me). :)
> Click up a terminal, and check with the commands:
Naturally, blinky owns both .fullcircle and .mozilla. And there is no
.firefox . Yes, every time I copy .mozilla in from otheruser, I chmod
it to blinky:blinky.
> firefox
> ls -al .fullcircle .mozilla .firefox
>
> If it still crashes, you might try this 5 command long command
>
> grep -i -r firefox .???* \
>| grep -i -v .fullcircle \
>| grep -i -v .mozilla \
>| grep -i -v .firefox \
>| grep -i -v .bash_history
>
> If nothing shows up, except for socket error messages, then I will have
This is way over my head. I'm sorry. Will you take a look? I can
learn from you, here.
http://blinkynet.net/stuff/comp/bash_history.txt
> to assume you have set some global value in your desktop manager which
> firefox does not like.
Perhaps I should open a Gnome session for blinky and try Ff there, then.
Does it make sense to try that before the following - which seems pretty
drastic, albeit certainly *possibly* necessary - suggestion?
> In that case I would logout of the desktop manager, and move the
> desktop manager directory (ies) and launch the desktop manager again.
> KDE Example:
> cd ~blinky
> pwd
> mv .kde .kde_broke
> Next login/launch of kde, under blinky, should create .kde and there you
> should be able to click up a terminal and run
> firefox
Also, I just had a thought. I did a backup of blinky's ~/ on the 23rd,
which was before FF started crashing on the 25th. That means I have a
copy of ~/.mozilla from before the crash. Now, I don't know that I
installed the new FF in the same directory -- would that be a problem
with trying blinky's old ~/.mozilla, or doesn't the stuff in ~/.mozilla
care about or refer to program location?
No complaints from me about your OP. :)
>> Right there, that tells you that your system installation works.
>> That assumes you do not have a local firefox install in that "other account"
>
> Right. It's in /bin/local .
>
>> No need to be screwing around with the system wide installed files.
>> It has to be something local to blinkey's account.
>
> Until I cleared everything out I didn't know that; I didn't think to
> look at another user before the great housecleaning. :)
Always assume the user account is dinked up.
As you saw, it can help to use an old established account for testing.
But to play safe, create a new test1 account to see if it works also. :)
I have a /junk/ account for testing in my custom setup and a /normal/ account
which runs without my custom changes in /etc/profile.d/*,
/etc/profile, /etc/bashrc.
> It still works for the other user, as noted in my OP. That has not
> changed since posting.
> 2.0.0.3. The only one now installed. I thought I noted that in OP.
I hear where you are commming from but I could not assume any order to
what you did or if lastest install was working. :-)
>> The thunderbird/firefox applications directories involved in /home/$USER are
>> .fullcircle, .thunderbird, .mozilla and .firefox
>
> I only have .mozilla (which contains firefox) and .fullcircle.
>
> Same as otheruser.
Good, old firefox releases has a .firefox. Just wanted to get you to a
known baseline for testshots.
> FF crashes on opening if .mozilla is deleted as stated in OP. It runs
> if otheruser's .mozilla is copied into Blinky's home dir -- but only
> until it encounters one of those death-dealing title bars.
Would help if you gave a url :(
> I can remove .fullcircle. That's safe?
Yep, do it all the time in bittwister's account when I want to test
something as a plain ole firefox user. I usually run firefox from
other accounts with a lots of pref/proxy settings.
>> That assumes you can create those directories in blinky's directory
>> and they are owned by blinky's UID/GID.
>
> .fullcircle has, as far as I know, always been there I've not done
> anything with it. I only noticed it when you started talking about it,
> since "fullcircle" does not exactly imply "firefox" (to me). :)
Yep, I create a new account, run an app and look around to see what's
new. :-)
Every once in awhile, I'll clean out .fullcircle.
>> Click up a terminal, and check with the commands:
>
> Naturally, blinky owns both .fullcircle and .mozilla. And there is no
> .firefox . Yes, every time I copy .mozilla in from otheruser, I chmod
> it to blinky:blinky.
Good, just wanted to point out that they have to be correct.
I do not know for sure that firefox is able to create the directories
in your setup. That is another reason I wanted you to delete them and
check them after firefox ran. :(
>> firefox
>> ls -al .fullcircle .mozilla .firefox
>>
>> If it still crashes, you might try this 5 command long command
>>
>> grep -i -r firefox .???* \
>>| grep -i -v .fullcircle \
>>| grep -i -v .mozilla \
>>| grep -i -v .firefox \
>>| grep -i -v .bash_history
>>
>> If nothing shows up, except for socket error messages, then I will have
>
> This is way over my head. I'm sorry. Will you take a look? I can
> learn from you, here.
>
> http://blinkynet.net/stuff/comp/bash_history.txt
Ok, next time you are not logged into kde under blinky, you can delete
all the ~blinky/.kde/socket* dir/files
While there, I would delete the .local/share/Trash/* stuff
and clean up the "No such file or directory" (guessing) links
I see nothing else at fault that is causing your FF problem.
>> to assume you have set some global value in your desktop manager which
>> firefox does not like.
>
> Perhaps I should open a Gnome session for blinky and try Ff there, then.
If gnome works and kde failes then that would indicate a kde global
setting to me.
> Does it make sense to try that before the following - which seems pretty
> drastic, albeit certainly *possibly* necessary - suggestion?
Hehe, to un-drastic the kde change, you just
/bin/rm -r .kde
mv .kde_broke .kde
and you are back as before. :)
Assuming you are not running kde.
> Also, I just had a thought. I did a backup of blinky's ~/ on the 23rd,
> which was before FF started crashing on the 25th. That means I have a
> copy of ~/.mozilla from before the crash. Now, I don't know that I
> installed the new FF in the same directory -- would that be a problem
> with trying blinky's old ~/.mozilla, or doesn't the stuff in ~/.mozilla
> care about or refer to program location?
Think about it for a minute. $HOME/.fullcircle and $HOME/.mozilla do
not exist for a new user. If the directories do not exists, then
the firefox script will create them.
That is why I suggested you delete them and let firefox create them.
That way I knew they would be created with blinky's UID/GID, 8-)
and have nothing in them which might be causing your problem.
You need to get to a known state before mixing in files from
backups/other locations.
As far as mozilla caring, I do not know for a fact, but experience has
shown me, so far, it does not care.
As you can see, my install has firefox in different locations
$ cd /local/opt
$ ls -al
drwxr-xr-x 13 root root 1024 Feb 24 13:14 firefox-2.0.0.2
drwxr-xr-x 13 root root 1024 Mar 21 16:24 firefox-2.0.0.3
drwxr-xr-x 10 root root 1024 Dec 7 02:05 thunderbird-1.5.0.9
$ type firefox
firefox is /usr/local/bin/firefox
$ ls -al /usr/local/bin/firefox
lrwxrwxrwx 1 root root 34 Mar 21 16:24 /usr/local/bin/firefox ->
/local/opt/firefox-2.0.0.3/firefox
I have yet to see a problem when I change the link in /usr/local/bin
to a new releases.
<big snip>
> Hehe, to un-drastic the kde change, you just
> /bin/rm -r .kde
> mv .kde_broke .kde
> and you are back as before. :)
> Assuming you are not running kde.
<big snip>
Dammit, I replied to this post at length, but it apparently didn't get
out. I must've got distracted by something I had to look at for our
discussion or someting I had to do for it, and not sent it before I
closed the session in which I was writing it. Dammit.
At any rate, here's what I've done from a root terminal:
Nuked .fullcircle and .mozilla, as you sugggested.
Renamed .kde to .kde_old. (But see 1, below!)
Then I started a KDE session as Blinky, ran Firefox for the first time
after that .kde housecleaning, and......found joy!
Thanks *so* much for sticking with me.
I'm happily surprised we - you - got it in in only three or four levels
of messages, here. After all the headbanging I'd done before I posted
this seemed like a huge obstacle.
I did lose all of my mail settings - folders, contents, filter rules -
the whole shebang - from Kmail. And all of my profile and server and
content and rules and filters - another whole shebang - from Knode,
which I do use just enough to have to rebuild it. I'm hoping I can find
at least some of that stuff from that backup I did of ~/ last Thursday
or so.
(1) That got ugly. I was in the root terminal trying to do this, and I
carelessly tried to rm it. I couldn't -- I got this error:
rm: cannot remove [filename]: Read-only file system
I got line after line of that. I'm root -- whaddaya mean I can't play
with some user files?! :)
So I closed the root terminal and went back to an otheruser Gnome
session, and everything I did there - run an app, whatever - just
normal stuff - I kept getting what errors about not having permission
to do that.
I got out of *that* session and rebooted, and was shown file system
error messages. That time I said no don't fix them. And then I
couldn't sign into a session as *anyone*. Wouldn't take anyone's
username/password. Not blinky, not otheruser, not root. I rebooted
again and fixed the files and then I got in okay. But that was scary
because last time I was locked out I had to (and was able to) go in as
root and fsck around with the user config. Since I couldn't even be
root this time, that didn't look too promising. Anyway, then I did
rename blinky's .kde as root from a terminal, as I'd started out to do
in the first place.
Ok, that explains that. I had seen your wireshark response in another post
and I waited up to 2:30am for your reply to this post.
Guessed the removal of 2 firefix dirs fixed the problem and I went to bed.
> At any rate, here's what I've done from a root terminal:
>
> Nuked .fullcircle and .mozilla, as you sugggested.
Did you run firefox to see if it still crashed?
> Renamed .kde to .kde_old. (But see 1, below!)
>
> Then I started a KDE session as Blinky, ran Firefox for the first time
> after that .kde housecleaning, and......found joy!
>
> I'm happily surprised we - you - got it in in only three or four levels
> of messages, here. After all the headbanging I'd done before I posted
> this seemed like a huge obstacle.
I would be nice to compare .kde and .kde_old files to see which
global setting(s) cause a url ad to crash firefox.
> I did lose all of my mail settings - folders, contents, filter rules -
> the whole shebang - from Kmail. And all of my profile and server and
> content and rules and filters - another whole shebang - from Knode,
> which I do use just enough to have to rebuild it. I'm hoping I can find
> at least some of that stuff from that backup I did of ~/ last Thursday
> or so.
Those settings are down in a .kde subdirectory. They could have been
copied from .kde_old into the same spot in .kde
You find them with
grep -r knode ~/.kde_old/*
grep -r kmail ~/.kde_old/*
Guessing they will be under ~/.kde_old/share/apps
> (1) That got ugly. I was in the root terminal trying to do this, and I
> carelessly tried to rm it. I couldn't -- I got this error:
>
> rm: cannot remove [filename]: Read-only file system
Please read http://www.catb.org/~esr/faqs/smart-questions.html
If we knew which [filename]: we might have a clue on why root was
having a problem.
> I got line after line of that. I'm root -- whaddaya mean I can't play
> with some user files?! :)
Without knowing the exact commands/sequence, there is no way to tell
you what went on.
Off hand, my SWAG would be you went to singel user runlevel and /home
is in a separate partition and it was not mounted.
> So I closed the root terminal and went back to an otheruser Gnome
> session, and everything I did there - run an app, whatever - just
> normal stuff - I kept getting what errors about not having permission
> to do that.
Well, that is wierd. Sounds like you were not in single user runlevel.
Did you do a
[_] su root
[_] su - root
[_] sudo something_here
Please check one of the above?
> I got out of *that* session and rebooted, and was shown file system
> error messages.
How did you reboot. Give commands and/or actions performed to reboot, please.
> That time I said no don't fix them. And then I
> couldn't sign into a session as *anyone*. Wouldn't take anyone's
> username/password. Not blinky, not otheruser, not root. I rebooted
> again and fixed the files and then I got in okay.
I thought Mandriva release 2007 automatically did a fsck on boot on a
dirty shutdonw.
What is your default runlevel. Show us the result from the following:
grep initdefault: /etc/inittab
> But that was scary because last time I was locked out I had to
> (and was able to) go in as root and fsck around with the user
> config. Since I couldn't even be root this time, that didn't look
> too promising. Anyway, then I did rename blinky's .kde as root from
> a terminal, as I'd started out to do in the first place.
Starting to guess we will see id:5:initdefault as the grep result of
/etc/inittab
If you were to have id:3:initdefault, blinky should have been able to
o logout of kde
o mv .kde .kde_old
o startx
o click up a terminal
o firefox
Downside of id:3:initdefault is, you have to enter startx every time
you login when you want a gui desktop manager.
As for your k*app settings, you can compare .kde and .kde_old file/dirs to
find/backup/cp files to .kde for your knode and whatnot.
Or, you can put some commands in your shell startup files
that do the startx for you under the proper conditions of
your choosing.
--
Robert Riches
spamt...@verizon.net
(Yes, that is one of my email addresses.)
> On 2007-03-27, Bit Twister <BitTw...@mouse-potato.com> wrote:
>>
>> Downside of id:3:initdefault is, you have to enter startx every time
>> you login when you want a gui desktop manager.
>
> Or, you can put some commands in your shell startup files
> that do the startx for you under the proper conditions of
> your choosing.
Like this from my .bash_profile:
print -n "Enter C for Console, Enter for X ==> "
read ans
if [ "$ans" != "C" -a "$ans" != 'c' ] ; then
exec startx
fi
I was up until 5:30 AM local time working on this and its ramifications
(see below <g>).
> Guessed the removal of 2 firefix dirs fixed the problem and I went to bed.
>> At any rate, here's what I've done from a root terminal:
>>
>> Nuked .fullcircle and .mozilla, as you sugggested.
>
> Did you run firefox to see if it still crashed?
I had about a hundred times with no .mozilla before. :) But no, I
didn't test it with .mozilla but without .fuulcircle. I just nuked them
both.
>> Renamed .kde to .kde_old. (But see 1, below!)
>>
>> Then I started a KDE session as Blinky, ran Firefox for the first time
>> after that .kde housecleaning, and......found joy!
>>
>> I'm happily surprised we - you - got it in in only three or four levels
>> of messages, here. After all the headbanging I'd done before I posted
>> this seemed like a huge obstacle.
>
> I would be nice to compare .kde and .kde_old files to see which
> global setting(s) cause a url ad to crash firefox.
It wasn't a URL. Again: anything that popped up a little window like
Help/About did it. That was my quick test for the problem all along.
>> I did lose all of my mail settings - folders, contents, filter rules -
>> the whole shebang - from Kmail. And all of my profile and server and
>> content and rules and filters - another whole shebang - from Knode,
>> which I do use just enough to have to rebuild it. I'm hoping I can find
>> at least some of that stuff from that backup I did of ~/ last Thursday
>> or so.
>
> Those settings are down in a .kde subdirectory. They could have been
> copied from .kde_old into the same spot in .kde
>
> You find them with
> grep -r knode ~/.kde_old/*
> grep -r kmail ~/.kde_old/*
> Guessing they will be under ~/.kde_old/share/apps
Yeah, I dealt with all that for hours. That's what kept me up. :)
But since I had that backup from a few days before the problem, I tried
pulling those from the BU DVD. I ran into all kinds of read-only files
in .kde/share/apps. I copied what I could, skipped the ones that
wouldn't, and neither Kmail or Knode is fixed. Knode (but not Kmail)
did show my old configuration stuff, right down to number of messages
read in each group. But when I tried to *use* it, it hung and after
several minutes I had to mercy-kill it. Kmail was still a fresh start
with no apparent old configuration stuff at all.
>> (1) That got ugly. I was in the root terminal trying to do this, and I
>> carelessly tried to rm it. I couldn't -- I got this error:
>>
>> rm: cannot remove [filename]: Read-only file system
>
> Please read http://www.catb.org/~esr/faqs/smart-questions.html
I link to it from blinkynet. I post links to it on Usenet.
> If we knew which [filename]: we might have a clue on why root was
> having a problem.
>
>> I got line after line of that. I'm root -- whaddaya mean I can't play
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
It was 4 AM and it was an effort to just read the fuzzy characters on
the screen, due to eyestrain, by then.
>> with some user files?! :)
>
> Without knowing the exact commands/sequence, there is no way to tell
> you what went on.
> Off hand, my SWAG would be you went to singel user runlevel and /home
> is in a separate partition and it was not mounted.
I was logged into otheruser/Gnome as only session . Then I F2'ed to a
terminal and logged in as root. That's where I tried to access the
/home/blinky/.kde stuff. Yes, I do have a separate partition for /home.
Good call. What a flippin' minefield. Well, then at least my *old*
.kde is intact, though. So when I get the small collateral fires put
out, I/we can see if we can figure out what in .kde was the culprit.
The small collateral fires include stuff that will no longer run. For
example, I'm still dialup, and I keep KDE's little net_applet utility
on my taskbar for connecting and disconnecting. I've added that to the
taskbar as usual, and while the button's there and pointing to the right
bin in /usr/bin and that file is present there, the button does nothing
when I right-click it to offer the list of actions (connect, disconnect,
monitor networks, configure networks, etc.). If I left click the button
I see net_applet's program button in the taskbar grinding away with the
hourglass, but it never starts; after thirty seconds or whatever the
setting is, that launch feedback button goes away and nothing has been
accomplished.
There was/were one or two other little things like that that Just Don't
Work now. I'll look for things in my software manager to (re)install,
but taking that net_applet for an example, its binary in /usr/bin is
a long way from ~/.kde, so I can't tell what's missing that's keeping
/usr/bin/net_applet from running. I wish I could remember which other
things seemed to be missing some parts, but I can't right now.
>> So I closed the root terminal and went back to an otheruser Gnome
>> session, and everything I did there - run an app, whatever - just
>> normal stuff - I kept getting what errors about not having permission
>> to do that.
In otheruser/Gnome. Yeah. I can't explain that a...er...Bit. :)
> Well, that is wierd. Sounds like you were not in single user runlevel.
> Did you do a
> [_] su root
> [_] su - root
> [_] sudo something_here
> Please check one of the above?
[x] None of the above.
Again:
1. I booted.
2. I opened an otheruser Gnome session.
3. I F2'ed to get to a terminal.
4. There, I logged in as root.
5. In that root terminal I tried to manipulate the files in
/home/blinky/.kde.
6. I got the read-only-file-system errors.
7. I closed the root session.
8. I F7'ed back to the otheruser/Gnome session.
9. Just about anything I tried to do there (none of which had anything
to do with /home/blinky - I just tried to do some normal
non-troubleshooting stuff in otheruser/Gnome) threw permissions
errors.
>> I got out of *that* session and rebooted, and was shown file system
>> error messages.
>
> How did you reboot. Give commands and/or actions performed to reboot, please.
The reboot that is #1, above: I don't remember if I rebooted through a
GUI or if I issued a halt or reboot from a root terminal. I *probably*
did it from a GUI.
I think the reboot I did after #9, above, was probably (which I know
doesn't help) a halt or reboot as root.
>> That time I said no don't fix them. And then I
>> couldn't sign into a session as *anyone*. Wouldn't take anyone's
>> username/password. Not blinky, not otheruser, not root. I rebooted
>> again and fixed the files and then I got in okay.
*That* reboot (the one that worked because I said Y to file fixes) was
started by the system when I said N to file fixes on the prior reboot.
> I thought Mandriva release 2007 automatically did a fsck on boot on a
> dirty shutdonw.
From OP: Mandraiva 2006
> What is your default runlevel. Show us the result from the following:
>
> grep initdefault: /etc/inittab
[blinky@thurston ~]$ grep initdefault: /etc/inittab
id:5:initdefault:
>> But that was scary because last time I was locked out I had to
>> (and was able to) go in as root and fsck around with the user
>> config. Since I couldn't even be root this time, that didn't look
>> too promising. Anyway, then I did rename blinky's .kde as root from
>> a terminal, as I'd started out to do in the first place.
>
> Starting to guess we will see id:5:initdefault as the grep result of
> /etc/inittab
Yes. It's not normally a problem.
> If you were to have id:3:initdefault, blinky should have been able to
> o logout of kde
> o mv .kde .kde_old
As whom?
> o startx
> o click up a terminal
> o firefox
>
> Downside of id:3:initdefault is, you have to enter startx every time
> you login when you want a gui desktop manager.
Probably why I've always (far's I know) used 5.
> As for your k*app settings, you can compare .kde and .kde_old file/dirs to
> find/backup/cp files to .kde for your knode and whatnot.
Perhaps. See above about my attempts to do that from DVD backup.
Yes. You can reduce false/nuisance instances of that
question by wrapping that block inside an if based on
whether or not the 'tty' command returns "not a tty" on the
value of (shell variable) $term, and on the value of
environment variable DISPLAY. The text virtual consoles
have a value for $term of 'linux', while terminal windows
running within X will have different values. $DISPLAY will
be set if the shell is running under X.
Note: all the following assumes no file corruption on disk or backup files.
> But since I had that backup from a few days before the problem, I tried
> pulling those from the BU DVD. I ran into all kinds of read-only files
> in .kde/share/apps. I copied what I could, skipped the ones that
> wouldn't, and neither Kmail or Knode is fixed. Knode (but not Kmail)
> did show my old configuration stuff, right down to number of messages
> read in each group. But when I tried to *use* it, it hung and after
> several minutes I had to mercy-kill it. Kmail was still a fresh start
> with no apparent old configuration stuff at all.
Then I have to guess the perms/UID/GID are not blinky's.
>> Please read http://www.catb.org/~esr/faqs/smart-questions.html
>
> I link to it from blinkynet. I post links to it on Usenet.
Yes, I have seen that, but I do not see actual messages/commands. :(
Except for the bash_log.txt :)
> I was logged into otheruser/Gnome as only session . Then I F2'ed to a
> terminal and logged in as root. That's where I tried to access the
> /home/blinky/.kde stuff. Yes, I do have a separate partition for /home.
> Good call. What a flippin' minefield. Well, then at least my *old*
> .kde is intact, though. So when I get the small collateral fires put
> out, I/we can see if we can figure out what in .kde was the culprit.
.kde_old/share/apps/ knode/kmail dirs aught to go back with no problem then.
As blinky
cd .kde_old/share/apps/
cp -a knode ~/.kde/share/apps
That should fix knode, if you are lucky.
> The small collateral fires include stuff that will no longer run. For
> example, I'm still dialup, and I keep KDE's little net_applet utility
> on my taskbar for connecting and disconnecting. I've added that to the
> taskbar as usual, and while the button's there and pointing to the right
> bin in /usr/bin and that file is present there, the button does nothing
> when I right-click it to offer the list of actions (connect, disconnect,
> monitor networks, configure networks, etc.). If I left click the button
> I see net_applet's program button in the taskbar grinding away with the
> hourglass, but it never starts; after thirty seconds or whatever the
> setting is, that launch feedback button goes away and nothing has been
> accomplished.
Guessing a ~/.kde/share/app/kppp setting lost when using the new .kde/* setup.
> Again:
>
> 1. I booted.
>
> 2. I opened an otheruser Gnome session.
>
> 3. I F2'ed to get to a terminal.
Ah so, I see, said the blind man.
I click a terminal, su - root
or just click my root keyboard shortcut script which pops me to a ssh logged in
root terminal.
No need for me to hassle with toggle f2/f7 activity.
Been trapped there, no desire to get caught again.
> 4. There, I logged in as root.
>
> 5. In that root terminal I tried to manipulate the files in
> /home/blinky/.kde.
>
> 6. I got the read-only-file-system errors.
Now, that makes this whole activity even more murky. What does
grep blinky /etc/passwd
grep home /etc/fstab
mount | grep home
have?
Also, I just want YOU to check uid/gid to verify they are correct on
any file/dirs. Assuming they BOTH are blinky, just verify please, no
need to post commands/results. 8-)
As blinky
click up a terminal,
ls -al and verify I have guessed the uid/gid correctly. If so just
run these commands and you should only see directory and total lines.
Anything else, needs to be corrected.
ls -al .kde/* | grep -v 'blinky blinky'
ls -al .kde/*/* | grep -v 'blinky blinky'
ls -al .kde/*/*/* | grep -v 'blinky blinky'
ls -al .kde/*/*/*/* | grep -v 'blinky blinky'
ls -al .kde/*/*/*/*/* | grep -v 'blinky blinky'
ls -al .kde/*/*/*/*/*/* | grep -v 'blinky blinky'
Do all those on ~/.kde_old and maybe the backup dvd.
> 7. I closed the root session.
>
> 8. I F7'ed back to the otheruser/Gnome session.
>
> 9. Just about anything I tried to do there (none of which had anything
> to do with /home/blinky - I just tried to do some normal
> non-troubleshooting stuff in otheruser/Gnome) threw permissions
> errors.
YOU might want to check your permissions/UID/GIDs. You may not want to
post yours, so here are mine.
$ ls -ald /home
drwxr-xr-x 5 root root 4096 Jan 5 19:50 /home
$ ls -al /home
drwxr-xr-x 5 root root 4096 Jan 5 19:50 .
drwxr-xr-x 36 root adm 4096 Mar 26 20:14 ..
drwx--x--x 63 bittwister bittwister 4096 Mar 27 15:14 bittwister
drwx--x--x 17 junk junk 4096 Mar 27 15:14 junk
drwx------ 16 normal normal 4096 Jan 22 04:07 normal
$ ls -ald .kde
drwx------ 7 bittwister bittwister 4096 Mar 27 12:52 .kde
>
>> What is your default runlevel. Show us the result from the following:
>>
>> grep initdefault: /etc/inittab
>
> [blinky@thurston ~]$ grep initdefault: /etc/inittab
> id:5:initdefault:
>
> Yes. It's not normally a problem.
Hehe, System acts differently depending on 5 or 3 and security level.
I have noticed people starting up in 5 have different results with
some kde settings than I do with starting up at 3.
>> If you were to have id:3:initdefault, blinky should have been able to
>> o logout of kde ^
>> o mv .kde .kde_old |
> |
> As whom?---------------------------------'
>> o startx
>> o click up a terminal
>> o firefox
>>
>> Downside of id:3:initdefault is, you have to enter startx every time
>> you login when you want a gui desktop manager.
>
> Probably why I've always (far's I know) used 5.
Until blinky's account gets dinked up somehow. 8-)
>> As for your k*app settings, you can compare .kde and .kde_old file/dirs to
>> find/backup/cp files to .kde for your knode and whatnot.
>
> Perhaps. See above about my attempts to do that from DVD backup.
Yeah, and your symptoms keep indicating perm/uid/gid problems of some
type on the target partition/user dir or from the source media.
Small snippet from my user dirs
$ ls -al .kde/*
.kde/Autostart:
drwxrwxr-x 2 bittwister bittwister 4096 Mar 2 08:00 .
drwx------ 7 bittwister bittwister 4096 Mar 27 12:52 ..
-rw-rw-r-- 1 bittwister bittwister 0 Jan 24 18:55 .alignment-icons
-rw-r--r-- 1 bittwister bittwister 1770 Jan 24 18:55 .directory
lrwxrwxrwx 1 bittwister bittwister 28 Mar 2 08:00 startup ->
/home/bittwister/bin/startup
.kde/share:
drwx------ 8 bittwister bittwister 4096 Mar 2 08:00 .
drwx------ 7 bittwister bittwister 4096 Mar 27 12:52 ..
drwx------ 2 bittwister bittwister 4096 Jan 24 18:55 applnk
drwx------ 26 bittwister bittwister 4096 Mar 12 12:54 apps
drwx------ 4 bittwister bittwister 4096 Mar 27 14:58 config
drwx------ 7 bittwister bittwister 4096 Jan 24 18:55 mimelnk
drwx------ 3 bittwister bittwister 4096 Jan 24 19:02 services
drwx------ 2 bittwister bittwister 4096 Feb 2 20:03 servicetypes
lrwxrwxrwx 1 bittwister bittwister 43 Mar 2 08:00 wallpapers ->
/accounts/bittwister/.kde/share/wallpapers/
$ ls -al .kde/share/apps
total 104
drwx------ 26 bittwister bittwister 4096 Mar 12 12:54 .
drwx------ 8 bittwister bittwister 4096 Mar 2 08:00 ..
drwx------ 2 bittwister bittwister 4096 Feb 10 17:21 digikam
drwx------ 2 bittwister bittwister 4096 Feb 23 22:44 k3b
drwx------ 2 bittwister bittwister 4096 Jan 24 18:55 kab
drwx------ 3 bittwister bittwister 4096 Jan 24 18:55 kabc
drwx------ 3 bittwister bittwister 4096 Jan 24 18:55 kconf_update
drwx------ 2 bittwister bittwister 4096 Mar 11 15:10 kcookiejar
drwx------ 2 bittwister bittwister 4096 Mar 12 12:54 kdeprint
drwx------ 4 bittwister bittwister 4096 Mar 27 05:32 kdesktop
> Then I have to guess the perms/UID/GID are not blinky's.
>>> Please read http://www.catb.org/~esr/faqs/smart-questions.html
>>
>> I link to it from blinkynet. I post links to it on Usenet.
>
> Yes, I have seen that, but I do not see actual messages/commands. :(
> Except for the bash_log.txt :)
I explained that lack.
>> I was logged into otheruser/Gnome as only session . Then I F2'ed to a
>> terminal and logged in as root. That's where I tried to access the
>> /home/blinky/.kde stuff. Yes, I do have a separate partition for /home.
>> Good call. What a flippin' minefield. Well, then at least my *old*
>> .kde is intact, though. So when I get the small collateral fires put
>> out, I/we can see if we can figure out what in .kde was the culprit.
>
> .kde_old/share/apps/ knode/kmail dirs aught to go back with no problem then.
>
> As blinky
> cd .kde_old/share/apps/
> cp -a knode ~/.kde/share/apps
>
> That should fix knode, if you are lucky.
I'll try.
>> The small collateral fires include stuff that will no longer run. For
>> example, I'm still dialup, and I keep KDE's little net_applet utility
>> on my taskbar for connecting and disconnecting. I've added that to the
>> taskbar as usual, and while the button's there and pointing to the right
>> bin in /usr/bin and that file is present there, the button does nothing
>> when I right-click it to offer the list of actions (connect, disconnect,
>> monitor networks, configure networks, etc.). If I left click the button
>> I see net_applet's program button in the taskbar grinding away with the
>> hourglass, but it never starts; after thirty seconds or whatever the
>> setting is, that launch feedback button goes away and nothing has been
>> accomplished.
>
> Guessing a ~/.kde/share/app/kppp setting lost when using the new .kde/* setup.
>
>> Again:
>>
>> 1. I booted.
>>
>> 2. I opened an otheruser Gnome session.
>>
>> 3. I F2'ed to get to a terminal.
>
> Ah so, I see, said the blind man.
> I click a terminal, su - root
I usually do that too. But with all that was going on, I wanted to
stick with the basics and opened a real root terminal.
> or just click my root keyboard shortcut script which pops me to a ssh logged in
> root terminal.
>
> No need for me to hassle with toggle f2/f7 activity.
> Been trapped there, no desire to get caught again.
>
>> 4. There, I logged in as root.
>>
>> 5. In that root terminal I tried to manipulate the files in
>> /home/blinky/.kde.
>>
>> 6. I got the read-only-file-system errors.
>
> Now, that makes this whole activity even more murky. What does
> grep blinky /etc/passwd
> grep home /etc/fstab
> mount | grep home
> have?
[blinky@thurston ~]$ grep blinky /etc/passwd
blinky:x:500:500:Blinky the Shark:/home/blinky:/bin/bash
[blinky@thurston ~]$ grep home /etc/fstab
/dev/hda6 /home ext3 defaults 1 2
[blinky@thurston ~]$ mount | grep home
/dev/hda6 on /home type ext3 (rw)
> Also, I just want YOU to check uid/gid to verify they are correct on
> any file/dirs. Assuming they BOTH are blinky, just verify please, no
> need to post commands/results. 8-)
> As blinky
> click up a terminal,
> ls -al and verify I have guessed the uid/gid correctly. If so just
> run these commands and you should only see directory and total lines.
> Anything else, needs to be corrected.
>
> ls -al .kde/* | grep -v 'blinky blinky'
.kde/Autostart:
total 16
.kde/DESKTOP_ENTRY:
total 8
.kde/share:
total 28
Which I believe means that like I said, everything's blinky:blinky.
For that level. As ls -al without the grep shows file by file.
> ls -al .kde/*/* | grep -v 'blinky blinky'
> ls -al .kde/*/*/* | grep -v 'blinky blinky'
> ls -al .kde/*/*/*/* | grep -v 'blinky blinky'
> ls -al .kde/*/*/*/*/* | grep -v 'blinky blinky'
> ls -al .kde/*/*/*/*/*/* | grep -v 'blinky blinky'
...and the same goes for all of those levels. Yes, I ran each one.
> Do all those on ~/.kde_old and maybe the backup dvd.
On the old dir, I got:
[blinky@thurston ~]$ ls -al .kde-old/*/* | grep -v 'blinky blinky'
srw------- 1 blinky nogroup 0 Mar 27 10:22 .kde-old/socket-localhost/kdesud_:0=
srw------- 1 blinky nogroup 0 Mar 27 10:22 .kde-old/socket-thurston.blinkynet.net/kdesud_:0=
But those were the only files that showed to be not blinky:blinky.
On the DVD, everything's root on the backup. But last night when copied
them into ~/.kde/share/apps I followed up with a chmod to blinky:blinky,
knowing that a whole directory of root-owned files in /home/blinky would
be pretty much worthless.
>> 7. I closed the root session.
>>
>> 8. I F7'ed back to the otheruser/Gnome session.
>>
>> 9. Just about anything I tried to do there (none of which had anything
>> to do with /home/blinky - I just tried to do some normal
>> non-troubleshooting stuff in otheruser/Gnome) threw permissions
>> errors.
>
> YOU might want to check your permissions/UID/GIDs. You may not want to
> post yours, so here are mine.
>
> $ ls -ald /home
> drwxr-xr-x 5 root root 4096 Jan 5 19:50 /home
[blinky@thurston blinky]$ ls -ald /home
drwxr-xr-x 5 root root 4096 Feb 25 2006 /home/
> $ ls -al /home
> drwxr-xr-x 5 root root 4096 Jan 5 19:50 .
> drwxr-xr-x 36 root adm 4096 Mar 26 20:14 ..
> drwx--x--x 63 bittwister bittwister 4096 Mar 27 15:14 bittwister
> drwx--x--x 17 junk junk 4096 Mar 27 15:14 junk
> drwx------ 16 normal normal 4096 Jan 22 04:07 normal
drwxr-xr-x 5 root root 4096 Feb 25 2006 ./
drwxr-xr-x 19 root root 4096 Mar 27 05:23 ../
drwxr-xr-x 86 blinky blinky 8192 Mar 27 14:15 blinky/
drwx------ 27 other other 4096 Mar 27 01:43 other/
drwxr-xr-x 2 root root 16384 Jan 12 2006 lost+found/
(I replaced the other user name with "other", above.)
> $ ls -ald .kde
> drwx------ 7 bittwister bittwister 4096 Mar 27 12:52 .kde
drwx------ 5 blinky blinky 4096 Mar 27 05:24 .kde/
>>> What is your default runlevel. Show us the result from the following:
>>>
>>> grep initdefault: /etc/inittab
>>
>> [blinky@thurston ~]$ grep initdefault: /etc/inittab
>> id:5:initdefault:
>>
>> Yes. It's not normally a problem.
>
> Hehe, System acts differently depending on 5 or 3 and security level.
> I have noticed people starting up in 5 have different results with
> some kde settings than I do with starting up at 3.
>
>
>>> If you were to have id:3:initdefault, blinky should have been able to
>>> o logout of kde ^
>>> o mv .kde .kde_old |
>> |
>> As whom?---------------------------------'
Yes, I know blinky was to log out of kde, but I didn't know if that mv
wsa supposed to be done by blinky or in a root session. Okay.
Yeah, I'm all blinky blinky.
su - root is a real root terminal login. :)
>
> [blinky@thurston ~]$ grep blinky /etc/passwd
> blinky:x:500:500:Blinky the Shark:/home/blinky:/bin/bash
>
> [blinky@thurston ~]$ grep home /etc/fstab
> /dev/hda6 /home ext3 defaults 1 2
>
> [blinky@thurston ~]$ mount | grep home
> /dev/hda6 on /home type ext3 (rw)
Well, for now, /home is read/write, but then again, you did reboot.
Next time you hit the perm problem, do a
whoami
pwd
ls -al
after root login and in whatever dir you are in.
>> ls -al .kde/*/*/*/*/*/* | grep -v 'blinky blinky'
>
> ...and the same goes for all of those levels. Yes, I ran each one.
>
>> Do all those on ~/.kde_old and maybe the backup dvd.
>
> On the old dir,
> these were the only files that showed to be not blinky:blinky.
>
> [blinky@thurston ~]$ ls -al .kde-old/*/* | grep -v 'blinky blinky'
> srw------- blinky nogroup .kde-old/socket-localhost/kdesud_:0=
> srw------- blinky nogroup .kde-old/socket-thurston.blinkynet.net/kdesud_:0=
Now, see there is a problem, why did they go nogroup. :(
.kde]$ ls -al
total 28
drwx------ 7 bittwister bittwister 4096 Mar 27 12:52 .
drwx--x--x 63 bittwister bittwister 4096 Mar 27 16:40 ..
drwxrwxr-x 2 bittwister bittwister 4096 Mar 2 08:00 Autostart
lrwxrwxrwx 1 bittwister bittwister 28 Mar 2 08:00 cache-wb.home.invalid ->
/var/tmp/kdecache-bittwister
lrwxrwxrwx 1 bittwister bittwister 39 Mar 2 08:00 socket-wb.home.invalid ->
/home/bittwister/tmp/ksocket-bittwister
lrwxrwxrwx 1 bittwister bittwister 35 Mar 2 08:00 tmp-wb.home.invalid ->
/home/bittwister/tmp/kde-bitt
>
>
> On the DVD, everything's root on the backup. But last night when copied
> them into ~/.kde/share/apps I followed up with a chmod to blinky:blinky,
> knowing that a whole directory of root-owned files in /home/blinky would
> be pretty much worthless.
I was guessing that, BUT you did not indicate chmod -R or what you ran. :(
>
>>> 7. I closed the root session.
>>>
>>> 8. I F7'ed back to the otheruser/Gnome session.
>>>
>>> 9. Just about anything I tried to do there (none of which had anything
>>> to do with /home/blinky - I just tried to do some normal
>>> non-troubleshooting stuff in otheruser/Gnome) threw permissions
>>> errors.
>>
>> YOU might want to check your permissions/UID/GIDs. You may not want to
>> post yours, so here are mine.
>>
>> $ ls -ald /home
>> drwxr-xr-x 5 root root 4096 Jan 5 19:50 /home
>
> [blinky@thurston blinky]$ ls -ald /home
>
1 > drwxr-xr-x 5 root root 4096 Feb 25 2006 ./
2 > drwxr-xr-x 19 root root 4096 Mar 27 05:23 ../
3 > drwxr-xr-x 86 blinky blinky 8192 Mar 27 14:15 blinky/
4 > drwx------ 27 other other 4096 Mar 27 01:43 other/
5 > drwxr-xr-x 2 root root 16384 Jan 12 2006 lost+found/
Looking at line 2 above, I wonder why it's GID is not adm.
I just booted my 2006 and it is adm. Looks like someone has been
playing with the chmod/chown -R command. :)
Not part of your problem, but I keep getting garbage information as to
where the problem should be. :(
>>>> If you were to have id:3:initdefault, blinky should have been able to
>>>> o logout of kde ^
>>>> o mv .kde .kde_old |
>>> |
>>> As whom?---------------------------------'
>
> Yes, I know blinky was to log out of kde, but I didn't know if that mv
> wsa supposed to be done by blinky or in a root session. Okay.
Only at runlevel 3 does the above work.
>> $ ls -al .kde/share/apps
>> total 104
>> drwx------ 26 bittwister bittwister 4096 Mar 12 12:54 .
>> drwx------ 2 bittwister bittwister 4096 Mar 12 12:54 kdeprint
>> drwx------ 4 bittwister bittwister 4096 Mar 27 05:32 kdesktop
>
> Yeah, I'm all blinky blinky.
Yes, but are/were all the /permissions/ correct? :(
My recommendation, assuming you have all updates installed, is to
continue to re-setup your k*apps and do not bring in any files from
old/bkups.
Keep notes of any global kde settings and addons, changes or whatnot.
> .kde_old/share/apps/ knode/kmail dirs aught to go back with no problem
> then.
>
> As blinky
> cd .kde_old/share/apps/
> cp -a knode ~/.kde/share/apps
>
> That should fix knode, if you are lucky.
The Latest News...
Replacing the new (bare-bones) .kde/share/apps/knode with the same directory
(recursively) from my save of the old (before we cleaned out .kde) and as
far as I can see I have everything back that I need for Knode. Note
headers. :)
Next up: Kmail.
Yeah, I'm being cautious.
Then yeah, I'll probably pull everything back into .../apps that I had
before. The new apps dir has a lot less stuff in it than the old one.
Dang, I did notice you using slrn and thought knode was fallback for
non-text groups. I was going to mention cleanscore which can erase
expired filters from your kill file.
>
> Next up: Kmail.
>
> Yeah, I'm being cautious.
>
> Then yeah, I'll probably pull everything back into .../apps that I had
> before. The new apps dir has a lot less stuff in it than the old one.
Yes, you get an apps dir when you fireup a kde app.
Note, I will recommend NOT running with the 2006 files when you goto
release 2007.
Line wrap caution:
http://qa.mandriva.com/twiki/bin/view/Main/MandrivaLinux2007Errata#UTF8_issue_when_reinstalling_and
> On Tue, 27 Mar 2007 16:53:12 -0800, Blinky the Shark wrote:
>>
>> Replacing the new (bare-bones) .kde/share/apps/knode with the same
>> directory (recursively) from my save of the old (before we cleaned out
>> .kde) and as
>> far as I can see I have everything back that I need for Knode. Note
>> headers. :)
>
> Dang, I did notice you using slrn and thought knode was fallback for
> non-text groups. I was going to mention cleanscore which can erase
> expired filters from your kill file.
No, it's for fun and learning and keeping on top of another client (I have
four up to speed in Linux (not counting Lynx <g>), and a three or four on
the Windows system). I just dig news clients; the first group I read
whenever I poll my main feed is news.software.readers. :) I also use
Knode with another identity with a real-sounding name for stuffy groups.
>> Next up: Kmail.
>>
>> Yeah, I'm being cautious.
>>
>> Then yeah, I'll probably pull everything back into .../apps that I had
>> before. The new apps dir has a lot less stuff in it than the old one.
>
> Yes, you get an apps dir when you fireup a kde app.
Discovered that today, with old and new .kde dirs open in two file manager
windows side by side, and running them in the order they appeared in the
old apps subdirectory. Watching them pop up on my left, in the new .kde
window. Some of them were already present there. I looked through all of
them, old and new, and if I saw info in the old that I though was relevant,
I (backed up the new) and moved it into the new. Everything seemed to
work.
Now, about Kmail. I did get it going and seeing my old list of folders and
the emails in those folders. That was phase one, accomplished when I went
through all of the apps directory. But even more was still missing --
identities, accounts, filters and address book. A TON of work to redo from
scratch, fershure by golly!
This prompted me to take at look at .kde/shared/config. And I found a
kmailrc in there. I had two old copies of .kde there, but since while
working on this project I'd gone ahead and set up a bare bones Kmail with
just one account, and no filters, they weren't what I needed. But I did
make that DVD backup of all of ~/ last week (the first I'd made in three
months, and I knew that was pushing my luck!), and, sure enough, I had just
the ticket there, it looked like. I copied it into shared/config, and
noticed the permissions were 444, inherited from the DVD. Not quite the
real deal when the program has to write to it, figured I, so I made it 600
like the rest of what was in that dir.
And except for identities, which I just recreated by hand, I seem to have a
fully regenerated Kmail.
I've also got most of my desktop set up like it was. :)
My eyes hurt.
> Note, I will recommend NOT running with the 2006 files when you goto
> release 2007.
Oh, great. :)
> Line wrap caution:
>
http://qa.mandriva.com/twiki/bin/view/Main/MandrivaLinux2007Errata#UTF8_issue_when_reinstalling_and
===
UTF8 issue when reinstalling and keeping a previous /home that was not in
UTF8
When installing instead of upgrading, and keeping a previous /home, the
install doesn't convert the /home to UTF8 and some characters might not
display properly in filenames.
To solve this issue, save the following convert-filenames-to-utf8.pl script
in your /home/UserName/, and run the following commands to convert
your /home/UserName/ to UTF8 :
cd ~/
chmod u+rx convert-filenames-to-utf8.pl
./convert-filenames-to-utf8.pl
===
Will save.
Even though I'm lookin' at it from a GUI session? I thought that
was some kinda emulation.
>> [blinky@thurston ~]$ grep blinky /etc/passwd
>> blinky:x:500:500:Blinky the Shark:/home/blinky:/bin/bash
>>
>> [blinky@thurston ~]$ grep home /etc/fstab
>> /dev/hda6 /home ext3 defaults 1 2
>>
>> [blinky@thurston ~]$ mount | grep home
>> /dev/hda6 on /home type ext3 (rw)
>
> Well, for now, /home is read/write, but then again, you did reboot.
>
> Next time you hit the perm problem, do a
> whoami
> pwd
> ls -al
> after root login and in whatever dir you are in.
Respectively:
root
/home/blinky , where I was when I su'ed.
I don't see anything surprising or special. What am I missing?
>>> ls -al .kde/*/*/*/*/*/* | grep -v 'blinky blinky'
>>
>> ...and the same goes for all of those levels. Yes, I ran each one.
>>
>>> Do all those on ~/.kde_old and maybe the backup dvd.
>>
>> On the old dir,
>> these were the only files that showed to be not blinky:blinky.
>>
>> [blinky@thurston ~]$ ls -al .kde-old/*/* | grep -v 'blinky blinky'
>> srw------- blinky nogroup .kde-old/socket-localhost/kdesud_:0=
>> srw------- blinky nogroup .kde-old/socket-thurston.blinkynet.net/kdesud_:0=
>
> Now, see there is a problem, why did they go nogroup. :(
Don't know, and never heard of it. :)
I've never used sudo, of that's what those are about.
> .kde]$ ls -al
> total 28
> drwx------ 7 bittwister bittwister 4096 Mar 27 12:52 .
> drwx--x--x 63 bittwister bittwister 4096 Mar 27 16:40 ..
> drwxrwxr-x 2 bittwister bittwister 4096 Mar 2 08:00 Autostart
> lrwxrwxrwx 1 bittwister bittwister 28 Mar 2 08:00 cache-wb.home.invalid ->
> /var/tmp/kdecache-bittwister
> lrwxrwxrwx 1 bittwister bittwister 39 Mar 2 08:00 socket-wb.home.invalid ->
> /home/bittwister/tmp/ksocket-bittwister
> lrwxrwxrwx 1 bittwister bittwister 35 Mar 2 08:00 tmp-wb.home.invalid ->
> /home/bittwister/tmp/kde-bitt
>
>>
>>
>> On the DVD, everything's root on the backup. But last night when copied
>> them into ~/.kde/share/apps I followed up with a chmod to blinky:blinky,
>> knowing that a whole directory of root-owned files in /home/blinky would
>> be pretty much worthless.
>
> I was guessing that, BUT you did not indicate chmod -R or what you ran. :(
I did it recursively. Not much good if only the top level's accessible.
By now, I don't know when that was run. But I did find today, and
reported it in another post, that stuff from the DVD came in 444 --
stuff I moved from .kde-old would've been okay.
> My recommendation, assuming you have all updates installed, is to
> continue to re-setup your k*apps and do not bring in any files from
> old/bkups.
I've never done an update, if you mean downloading updates to Mandriva.
Started with I think I came aboard at 8.0 (might've been 8.2). Dialup;
I don't do huge downloads. Hell, I can't even get the packages managers
to see at the RPMs in my download directory, much less try to get it to
see something offsite. So I just use urpmi and if I don't have
something here I don't install it.
> I copied it into shared/config, and
> noticed the permissions were 444, inherited from the DVD. Not quite the
> real deal when the program has to write to it, figured I, so I made it 600
> like the rest of what was in that dir.
Yes, those are the kinds of problems I would expect and had assumed
were most of your problems when files came from backup.
>> Note, I will recommend NOT running with the 2006 files when you goto
>> release 2007.
>
> Oh, great. :)
>
>> Line wrap caution:
>>
> http://qa.mandriva.com/twiki/bin/view/Main/MandrivaLinux2007Errata#UTF8_issue_when_reinstalling_and
>
> ===
>
> UTF8 issue when reinstalling and keeping a previous /home that was not in
> UTF8
>
> When installing instead of upgrading, and keeping a previous /home, the
> install doesn't convert the /home to UTF8 and some characters might not
> display properly in filenames.
I remember once when kde changed the format of the desktop
shortcuts. I wondered why .xsession-errors file was so big. Nothing
but complaints about my desktop shortcuts. I deleted all and recreated them
and problem cleared up.
It is little gotchas like that which keep me from using some of the
features of the apps.
Take email and bookmarks.
I keep those in a separate ascii files. Now it is easy to use grep to find
what I want and it is not hard at all to setup accounts and know there
will not be a conflict in the file format.
As a matter of fact, it is easier to use.
I wrote a script which allows me to use _keywords_ to help find them faster.
Urls example:
$ urls bash doc
http://gentoo-wiki.com/MAN_bash ! documentation
http://tldp.org/LDP/abs/html/index.html ! bash script advanced documentation
http://tldp.org/LDP/intro-linux/html/index.html ! bash script introduction
to Linux documentation, see IO redirection
Another example
$ urls java test
http://java.com/en/download/installed.jsp ! test java virtual machine install
http://www.bodo.com/javame.htm ! test java
http://www.hedonistica.com ! java test games
Anytime someone posts a url which I think I might need, I use my alias
"eurls" which pop me into editing the urls file where I add the url
and some keywords to help find it later.
I made the script ambidexterous for searching apropos, spelling dictionary,
services, mirror index files, /var/log/messages, locate result, ...
#******* start of ux ****************************************
#*
#* ux - search Unix Help, urls, email files with keywords.
#*
#* basename is called to determine command and file to use.
#*
#* To add a new selection:
#* copy a case $_exe stanza, set case_name and _tmp
#* create a link with ln -s ux new_case_name
#*
#* Examples: ls -l u*
#* ue -> ux
#* uh -> ux
#* uidx -> ux
#* ulocate -> ux
#* uman -> ux
#* userv -> ux
#* urgb -> ux
#* urls -> ux
#* userv -> ux
#* and to find a word in dictionary there is
#* sp -> ux
#* sp usage example would be sp amb d x to find ambidexterous
#*
#*
#**********************************************************
_tmp_fn=$HOME/tmp/ux
_cnt=$#
a1=$1
a2=$2
a3=$3
a4=$4
a5=$5
a6=$6
_exe=$(basename $0)
if [ ! -e $HOME/tmp ] ; then
mkdir $HOME/tmp
fi
# tput clear
echo "echo -en '\033[?3h\033[?66h'" > $_tmp_fn
case $_cnt in
1)
_tmp=" "
;;
2)
_tmp=" | grep -i $a2"
;;
3)
_tmp=" | grep -i $a2 | grep -i $a3"
;;
4)
_tmp=" | grep -i $a2 | grep -i $a3 | grep -i $a4"
;;
5)
_tmp=" | grep -i $a2 | grep -i $a3 | grep -i $a4 | grep -i $a5"
;;
6)
_tmp=" | grep -i $a2 | grep -i $a3 | grep -i $a4 | grep -i $a5 | grep -i $a6"
;;
*)
echo "echo usage: $_exe keyword1 [keyword2] [keyword3] [keyword4] [keyword5] [keyword6]" >> $_tmp_fn
;;
esac
case $_exe in
ue)
echo "grep -i $a1 /accounts/bittwister/bin/send_humor.pl $_tmp" >> $_tmp_fn
;;
uh)
echo "grep -i $a1 /site/doc/unix.help $_tmp" >> $_tmp_fn
;;
ui)
echo "grep -i $a1 /site/doc/install.changes.${_release} $_tmp" >> $_tmp_fn
;;
ulocate)
echo "locate -i $a1 $_tmp" >> $_tmp_fn
;;
uidx)
echo "cd /local/spare/" >> $_tmp_fn
echo "grep -i $a1 *idx* $_tmp" >> $_tmp_fn
;;
uman)
echo "apropos $a1 $_tmp" >> $_tmp_fn
;;
umessage)
echo "grep -i $a1 /var/log/messages $_tmp" >> $_tmp_fn
;;
userv)
echo "grep -i $a1 /etc/services $_tmp" >> $_tmp_fn
;;
urls)
echo "grep -i $a1 /site/doc/urls $_tmp" >> $_tmp_fn
;;
urgb)
echo "showrgb | grep -i $a1 $_tmp" >> $_tmp_fn
;;
sp)
echo "look $a1 /usr/share/dict/words $_tmp" >> $_tmp_fn
;;
*)
echo "Unknown exe name $_exe" >> $_tmp_fn
exit 1
;;
esac
echo /bin/rm $_tmp_fn >> $_tmp_fn
chmod +x $_tmp_fn
$_tmp_fn
#********************* end ux ******************************
At best, you are talking terminal emulation which has nothing to do
with who is logged in, in the terminal.
>>
>> Next time you hit the perm problem, do a
>> whoami
>> pwd
>> ls -al
>> after root login and in whatever dir you are in.
>
> Respectively:
>
> root
>
> /home/blinky , where I was when I su'ed.
Augh, frap, You said you F2'ed not su'ed :( 8-)
Now, those instructions were for when you encounter permission
problems in the future when running as root.
/su root/ and /su - root/ do not do the same thing.
>>>
>>> [blinky@thurston ~]$ ls -al .kde-old/*/* | grep -v 'blinky blinky'
>>> srw------- blinky nogroup .kde-old/socket-localhost/kdesud_:0=
>>> srw------- blinky nogroup .kde-old/socket-thurston.blinkynet.net/kdesud_:0=
>>
>> Now, see there is a problem, why did they go nogroup. :(
>
> Don't know, and never heard of it. :)
I was just pointing out you had some kind of glitch and somehow they
lost their GID. Dinking with firefox addons would/should not cause
that problem.
>> I was guessing that, BUT you did not indicate chmod -R or what you ran. :(
>
> I did it recursively. Not much good if only the top level's accessible.
Always keep in mind, I try to keep the lurkers in mind when replying
to posts and I was not shown what you did.
When debugging a problem, we need to see what was done, not hear about
what you said you did. :)
> I've never done an update, if you mean downloading updates to Mandriva.
> Started with I think I came aboard at 8.0 (might've been 8.2). Dialup;
> I don't do huge downloads. Hell, I can't even get the packages managers
> to see at the RPMs in my download directory, much less try to get it to
> see something offsite. So I just use urpmi and if I don't have
> something here I don't install it.
Well, there are all sorts of problems being fixed all the time.
Matter of fact, a few kde problems.
You might want to look at
http://www.mandriva.com/security/advisories?dis=2006.0
My neighbor is on dial-up and I understand your problem. I set her
mirrors, set update boxes, disabled removable media, and created a
script for her to run.
Now she boots up the system every night before going to bed, runs kppp
to open the connection, clicks a desktop shortcut to bring up a root
terminal, runs getupdates, power off the monitor and goes to bed.
Updates are installed and system powers off when done.
One time, there was an office update and a few others and it was still running
downloads the next morning.
$ cat getupdates
urpmi --wget --update --auto-update --auto
poweroff
Okay.
>>> Next time you hit the perm problem, do a
>>> whoami
>>> pwd
>>> ls -al
>>> after root login and in whatever dir you are in.
>>
>> Respectively:
>>
>> root
>>
>> /home/blinky , where I was when I su'ed.
>
> Augh, frap, You said you F2'ed not su'ed :( 8-)
Might've been either. After three days of all this, and probably 90
changes of user, I certainly can't remember now. I tried to be as
accurate as possible at the time.
> Now, those instructions were for when you encounter permission
> problems in the future when running as root.
>
> /su root/ and /su - root/ do not do the same thing.
I'll look up "su - root". I'm not aware of that command.
>>>>
>>>> [blinky@thurston ~]$ ls -al .kde-old/*/* | grep -v 'blinky blinky'
>>>> srw------- blinky nogroup .kde-old/socket-localhost/kdesud_:0=
>>>> srw------- blinky nogroup .kde-old/socket-thurston.blinkynet.net/kdesud_:0=
>>>
>>> Now, see there is a problem, why did they go nogroup. :(
>>
>> Don't know, and never heard of it. :)
>
> I was just pointing out you had some kind of glitch and somehow they
> lost their GID. Dinking with firefox addons would/should not cause
> that problem.
Woldn't think so, no.
>>> I was guessing that, BUT you did not indicate chmod -R or what you ran. :(
>>
>> I did it recursively. Not much good if only the top level's accessible.
>
> Always keep in mind, I try to keep the lurkers in mind when replying
> to posts and I was not shown what you did.
Thanks. That means I can let some stuff to without comment.
> When debugging a problem, we need to see what was done, not hear about
> what you said you did. :)
Right now, I'm the best reporter on site. But come on over. Like to
fly? ;)
>> I've never done an update, if you mean downloading updates to Mandriva.
>> Started with I think I came aboard at 8.0 (might've been 8.2). Dialup;
>> I don't do huge downloads. Hell, I can't even get the packages managers
>> to see at the RPMs in my download directory, much less try to get it to
>> see something offsite. So I just use urpmi and if I don't have
>> something here I don't install it.
>
> Well, there are all sorts of problems being fixed all the time.
> Matter of fact, a few kde problems.
>
> You might want to look at
> http://www.mandriva.com/security/advisories?dis=2006.0
>
> My neighbor is on dial-up and I understand your problem. I set her
> mirrors, set update boxes, disabled removable media, and created a
> script for her to run.
>
> Now she boots up the system every night before going to bed, runs kppp
> to open the connection, clicks a desktop shortcut to bring up a root
> terminal, runs getupdates, power off the monitor and goes to bed.
> Updates are installed and system powers off when done.
What happens when all this hits dependency hell and grinds to a halt
until she gets up, puts on her nice clothes and makeup and shuts it down
in frustration before leaving for work? Oh, wait...*that* happens? ;)
> One time, there was an office update and a few others and it was still running
> downloads the next morning.
:)
Send me the round trip ticket, have a rental car with credit
card and map to your house in the glove box at the airport and it's a
deal. 8-)
> What happens when all this hits dependency hell and grinds to a halt
> until she gets up,
I have yet to see that problem. Then again, I have loaded all package
groups except LSB, both gnome and kde desktop managers and checks for
updates are daily.
I have set firefox and thunderbird for internet work.
I have shown here how to check the kppp connection before calling me
if it is still running.
> puts on her nice clothes and makeup and shuts it down
> in frustration before leaving for work? Oh, wait...*that* happens? ;)
At worst, I would get a phone call, walk next door, and see what happened.
Other times she reads me the screen over the phone and I do not have
to take the walk.
Of course, since I do my updates via cable modem, before she goes to
bed, I have a pretty good idea what will happen at night on her pc. 8-)
We have the same hardware and mirror settings.
We both had kde problems until the updates came out.
Been pretty solid since.
Sorry. I'm only good for the beer and peanuts. :)
>> What happens when all this hits dependency hell and grinds to a halt
>> until she gets up,
>
> I have yet to see that problem. Then again, I have loaded all package
> groups except LSB, both gnome and kde desktop managers and checks for
> updates are daily.
>
> I have set firefox and thunderbird for internet work.
>
> I have shown here how to check the kppp connection before calling me
> if it is still running.
>
>> puts on her nice clothes and makeup and shuts it down
>> in frustration before leaving for work? Oh, wait...*that* happens? ;)
>
> At worst, I would get a phone call, walk next door, and see what happened.
> Other times she reads me the screen over the phone and I do not have
> to take the walk.
>
> Of course, since I do my updates via cable modem, before she goes to
> bed, I have a pretty good idea what will happen at night on her pc. 8-)
Well, they're not *identically* set up. You might still get a few
surprises.
> We have the same hardware and mirror settings.
Okay, cancel what I just said. :)
> We both had kde problems until the updates came out.
> Been pretty solid since.
Great.