LDM crashing on trying to log in.

Discussion:

Gary Nutbeam

2013-10-14 15:03:02 UTC

I'm having stability problems with LDM. I'm using decent thin clients
now (1700 model from disklessworkstations.com, hyperthreading atom, 2GB
RAM etc.).

I'm using ltsp-server 5.4.5-24 on CentOS 6.4 64bit from the ltsp repo at
ltsprepo.s3.amazonaws.com.

When first attempting to log in, after entering the username, LDM seems
to crash, and restart. The second attempt to login works fine. If it
doesn't crash on the first attempt, the X session crashes very shortly
afterwards, usually when trying to launch an application. After the
initial weirdness, everything seems to settle down and work.

This behavior happens on both the physical thin clients AND a virtual
client.

Has anyone seen this problem and know what might be causing it?

Thanks.

Gary.

k12ltsp

2013-10-14 15:42:37 UTC

Permalink

Hi Gary,

We have the same problem I'm afraid.
Using ltsp-server-5.4.5-24.elg.x86_64 but with 32 bit clients. Atom
processors in the clients with at least 1GByte RAM.
Usually work reliably once logged in.

Also cannot shutdown or reboot the client. Tried some suggestions from
the list, but they didn't help.

Things seem to have got worse recently, and I don't have the skills to
try and fix it, we do IT support for small businesses and web/database apps.

Now trying Debian Wheezy with LTSP - just works, so looks like I shall
jump ship.

Regards
Trevor
http://www.infocentrality.co.uk

Post by Gary Nutbeam
I'm having stability problems with LDM. I'm using decent thin clients
now (1700 model from disklessworkstations.com, hyperthreading atom, 2GB
RAM etc.).
I'm using ltsp-server 5.4.5-24 on CentOS 6.4 64bit from the ltsp repo at
ltsprepo.s3.amazonaws.com.
When first attempting to log in, after entering the username, LDM seems
to crash, and restart. The second attempt to login works fine. If it
doesn't crash on the first attempt, the X session crashes very shortly
afterwards, usually when trying to launch an application. After the
initial weirdness, everything seems to settle down and work.
This behavior happens on both the physical thin clients AND a virtual
client.
Has anyone seen this problem and know what might be causing it?
Thanks.
Gary.
_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

Gary Nutbeam

2013-10-14 16:02:33 UTC

Permalink

It would be a lot of work for me to move to Debian, because of
applications to be moved, tested on Debian, etc. Plus everything we do
is CentOS and Fedora based.

I can't shutdown or reboot the client either. I saw a hack on the list
that I haven't tried yet to "fix" the shutdown/reboot problem.

Post by k12ltsp
Hi Gary,
We have the same problem I'm afraid.
Using ltsp-server-5.4.5-24.elg.x86_64 but with 32 bit clients. Atom
processors in the clients with at least 1GByte RAM.
Usually work reliably once logged in.
Also cannot shutdown or reboot the client. Tried some suggestions from
the list, but they didn't help.
Things seem to have got worse recently, and I don't have the skills to
try and fix it, we do IT support for small businesses and web/database apps.
Now trying Debian Wheezy with LTSP - just works, so looks like I shall
jump ship.
Regards
Trevor
http://www.infocentrality.co.uk

_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

Joshua Trimm

2013-10-14 22:03:20 UTC

Permalink

The EPEL repo is the current most stable one, I have had some issues
with the server that hosts my development code. It had to do a lost file
recovery so i'm thinking a beta version got into the repo. I'll fix when
I can, but for now use the rpms from EPEL.

Post by Gary Nutbeam
It would be a lot of work for me to move to Debian, because of
applications to be moved, tested on Debian, etc. Plus everything we do
is CentOS and Fedora based.
I can't shutdown or reboot the client either. I saw a hack on the list
that I haven't tried yet to "fix" the shutdown/reboot problem.

Post by k12ltsp
Hi Gary,
We have the same problem I'm afraid.
Using ltsp-server-5.4.5-24.elg.x86_64 but with 32 bit clients. Atom
processors in the clients with at least 1GByte RAM.
Usually work reliably once logged in.
Also cannot shutdown or reboot the client. Tried some suggestions from
the list, but they didn't help.
Things seem to have got worse recently, and I don't have the skills to
try and fix it, we do IT support for small businesses and
web/database apps.
Now trying Debian Wheezy with LTSP - just works, so looks like I shall
jump ship.
Regards
Trevor
http://www.infocentrality.co.uk

_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

Burke Almquist

2013-10-15 06:26:02 UTC

Permalink

The EPEL repo is the current most stable one, I have had some issues with the server that hosts my development code. It had to do a lost file recovery so i'm thinking a beta version got into the repo. I'll fix when I can, but for now use the rpms from EPEL.

I've noticed the same thing on Scientific Linux 6.4. I went back to using the "old" install instructions on the wiki, the ones that use the EPEL repos. I think the ones on amazon S3 are maybe only usable on Fedora 18? Why they are mixing them up with the EL install instructions is beyond me. Trying to build using the new instructions a couple months ago it complained about not having the Mock package and a certain version installed. No one seemed to know why the instructions were changed so I just went back to the old ones. Seems to be working for the most part.

Gary Nutbeam

2013-10-15 17:35:33 UTC

Permalink

I shall do that. Thanks.

Post by Joshua Trimm
The EPEL repo is the current most stable one, I have had some issues
with the server that hosts my development code. It had to do a lost file
recovery so i'm thinking a beta version got into the repo. I'll fix when
I can, but for now use the rpms from EPEL.

Post by k12ltsp
Hi Gary,
We have the same problem I'm afraid.
Using ltsp-server-5.4.5-24.elg.x86_64 but with 32 bit clients. Atom
processors in the clients with at least 1GByte RAM.
Usually work reliably once logged in.
Also cannot shutdown or reboot the client. Tried some suggestions from
the list, but they didn't help.
Things seem to have got worse recently, and I don't have the skills to
try and fix it, we do IT support for small businesses and
web/database apps.
Now trying Debian Wheezy with LTSP - just works, so looks like I shall
jump ship.
Regards
Trevor
http://www.infocentrality.co.uk

_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

Peter Onion

2013-10-18 11:36:50 UTC

Permalink

Post by Gary Nutbeam
I can't shutdown or reboot the client either. I saw a hack on the list
that I haven't tried yet to "fix" the shutdown/reboot problem.

I found there are issues with some of the scripts in /etc/init.d trying
to stop the networking during shutdown, which is NOT a good idea when
the root file system is mounted on NFS !

ISTR I added "exit 1" at the top of the "stop" sections in 3 or 4 files
but I don't have access to a working example at the moment so I can't
give you a list until next week.

PeterO

Gary Nutbeam

2013-10-18 15:57:09 UTC

Permalink

That would be much appreciated.

Post by Peter Onion

Post by Gary Nutbeam
I can't shutdown or reboot the client either. I saw a hack on the list
that I haven't tried yet to "fix" the shutdown/reboot problem.

I found there are issues with some of the scripts in /etc/init.d trying
to stop the networking during shutdown, which is NOT a good idea when
the root file system is mounted on NFS !
ISTR I added "exit 1" at the top of the "stop" sections in 3 or 4 files
but I don't have access to a working example at the moment so I can't
give you a list until next week.
PeterO
_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

Radek Bursztynowski

2013-10-14 21:51:42 UTC

Permalink

Hello,

I have the same problems. The problem with first attempting to log in I
solved (I believe). This solution concerns i386 and x86_64 thin client
image (epel6-i386 and epel-6-x86_64).

I added to /opt/ltsp/chroot/etc/sysconfig three files which I couldn't
find after ltsp-build-client script execution:

(Polish parameters)
clock
ZONE="Warsaw/Europe"

i18n
LANG="pl_PL.UTF-8"

keyboard
KEYTABLE="pl2"
MODEL="pc105"
LAYOUT="pl"
KEYBOARDTYPE="pc"

and my /opt/ltsp/i386/etc/ltsp_chroot is not empty but includes

LTSP_CHROOT=/opt/ltsp/i386 line

and /opt/ltsp/x86_64/etc/ltsp_chroot includes

LTSP_CHROOT=/opt/ltsp/x86_64 line.

Now LDM doesn't reload and works fine.

I am not sure that this the best solution, but it helped me.

Thin client still doesn't reboot or shutdown.

I tried Debian LTSP too, but I retrieved. Finally I moved Debian thin
client image to LTSP on CentOS 6.4 and I can use it, but I prefer epel-6
images. In my opinion the most stable thin client image is Scientific
Linux 6.1 (ltsp-server 5.2) and I recommend it.

Best regards,
Radek

---

_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

Gary Nutbeam

2013-10-15 15:39:08 UTC

Permalink

Thanks for the info. Unfortunately I tried this with US information
substituted and still have LDM crashing. I was running a Scientific
Linux client and it was working better. I'll see if I can rollback to
that, or maybe try the version out of the EPEL repo.

Post by Radek Bursztynowski
Hello,
I have the same problems. The problem with first attempting to log in I
solved (I believe). This solution concerns i386 and x86_64 thin client
image (epel6-i386 and epel-6-x86_64).
I added to /opt/ltsp/chroot/etc/sysconfig three files which I couldn't
(Polish parameters)
clock
ZONE="Warsaw/Europe"
i18n
LANG="pl_PL.UTF-8"
keyboard
KEYTABLE="pl2"
MODEL="pc105"
LAYOUT="pl"
KEYBOARDTYPE="pc"
and my /opt/ltsp/i386/etc/ltsp_chroot is not empty but includes
LTSP_CHROOT=/opt/ltsp/i386 line
and /opt/ltsp/x86_64/etc/ltsp_chroot includes
LTSP_CHROOT=/opt/ltsp/x86_64 line.
Now LDM doesn't reload and works fine.
I am not sure that this the best solution, but it helped me.
Thin client still doesn't reboot or shutdown.
I tried Debian LTSP too, but I retrieved. Finally I moved Debian thin
client image to LTSP on CentOS 6.4 and I can use it, but I prefer epel-6
images. In my opinion the most stable thin client image is Scientific
Linux 6.1 (ltsp-server 5.2) and I recommend it.
Best regards,
Radek
---

_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

Radek Bursztynowski

2013-10-18 14:23:57 UTC

Permalink

I tested it the once again. I can confirm that trying to log in that
epel-6-i386 image installed 2 mounts ago crash on trying to log in.

But the image installed few days ago using my parameters works fine. I
copied this to another one LTSP server, and I have just finish the test
of copied image. Works fine with no crashing on trying to log in.

If you want to try my working image you can download it, both i386 and
x86_64 versions:
http://www.bursztynowski.pl/ftp/linux/LinuxTerminalServerProject/5/5.4/epel-6-thin_client_image/

I renamed chroot directory to epel-6-i386_base and epel-6-x86_64_base.

Best regards,
Radek

---

Post by Gary Nutbeam
Thanks for the info. Unfortunately I tried this with US information
substituted and still have LDM crashing. I was running a Scientific
Linux client and it was working better. I'll see if I can rollback to
that, or maybe try the version out of the EPEL repo.

_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

Gary Nutbeam

2013-10-18 15:57:24 UTC

Permalink

Thanks. I'm going to download the image and have a look at it.

Post by Radek Bursztynowski
I tested it the once again. I can confirm that trying to log in that
epel-6-i386 image installed 2 mounts ago crash on trying to log in.
But the image installed few days ago using my parameters works fine. I
copied this to another one LTSP server, and I have just finish the test
of copied image. Works fine with no crashing on trying to log in.
If you want to try my working image you can download it, both i386 and
http://www.bursztynowski.pl/ftp/linux/LinuxTerminalServerProject/5/5.4/epel-6-thin_client_image/
I renamed chroot directory to epel-6-i386_base and epel-6-x86_64_base.
Best regards,
Radek
---

_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>