Solution for singularity error: GLIBC_2.34 not found (Ubuntu22.04)

dbpatankar

Hi,

I have installed salome_meca 2021 container image using singularity on my arco-linux machine. So far it worked without any issue and I was able to use it. But now may be after an upgrade the salome_meca is unable to start. Here follows what happens and what I know about the issue :

1. When I run it in GPU mode, I get the GLIBC_2.34 error message

$./salome_meca-lgpl-2021.0.0-2-20211014-scibian-9
/usr/bin/nvidia-smi
**************************************
INFO : Running salome_meca in GPU mode
**************************************
runSalome running on digvijay-gl552vx
Searching for a free port for naming service: 2812 - OK
Searching Naming Service  + found in 0.1 seconds 
Searching /Kernel/Session in Naming Service  ++++SALOME_Session_Server: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.34' not found (required by /.singularity.d/libs/libGLX.so.0)
Traceback (most recent call last):
  File "/opt/salome_meca/Salome-V2021-s9/modules/KERNEL_V9_7_0/bin/salome/orbmodule.py", line 181, in waitNSPID
    os.kill(thePID,0)
ProcessLookupError: [Errno 3] No such process

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/opt/salome_meca/appli_V2021/bin/salome/runSalome.py", line 694, in useSalome
    clt = startSalome(args, modules_list, modules_root_dir)
  File "/opt/salome_meca/appli_V2021/bin/salome/runSalome.py", line 639, in startSalome
    session=clt.waitNSPID("/Kernel/Session",mySessionServ.PID,SALOME.Session)
  File "/opt/salome_meca/Salome-V2021-s9/modules/KERNEL_V9_7_0/bin/salome/orbmodule.py", line 183, in waitNSPID
    raise RuntimeError("Process %d for %s not found" % (thePID,theName))
RuntimeError: Process 8691 for /Kernel/Session not found


--- Error during Salome launch ---

2. When I run it in software rendering mode, it starts properly though rendering is not proper:

$./salome_meca-lgpl-2021.0.0-2-20211014-scibian-9 --soft ***************************************************** INFO : Running salome_meca in software rendering mode ***************************************************** runSalome running on digvijay-gl552vx Searching for a free port for naming service: 2813 - OK Searching Naming Service + found in 0.1 seconds Searching /Kernel/Session in Naming Service +++++++++++ found in 5.5 seconds Start SALOME, elapsed time : 6.1 seconds **************************************************************** Warning: module GHS3DPLUGIN is improperly configured! Module GHS3DPLUGIN will not be available in GUI mode! **************************************************************** **************************************************************** Warning: module GHS3DPRLPLUGIN is improperly configured! Module GHS3DPRLPLUGIN will not be available in GUI mode! **************************************************************** **************************************************************** Warning: module BLSURFPLUGIN is improperly configured! Module BLSURFPLUGIN will not be available in GUI mode! **************************************************************** **************************************************************** Warning: module NETGENPLUGIN is improperly configured! Module NETGENPLUGIN will not be available in GUI mode! **************************************************************** **************************************************************** Warning: module HYBRIDPLUGIN is improperly configured! Module HYBRIDPLUGIN will not be available in GUI mode! **************************************************************** **************************************************************** Warning: module GMSHPLUGIN is improperly configured! Module GMSHPLUGIN will not be available in GUI mode! **************************************************************** **************************************************************** Warning: module HexoticPLUGIN is improperly configured! Module HexoticPLUGIN will not be available in GUI mode!

****************************************************************

3. I can get into the container shell and inspect the environment:

Singularity> ldd --version ldd (Debian GLIBC 2.24-11+deb9u4) 2.24 Copyright © 2016 Free Software Foundation, Inc. Ce logiciel est libre; voir les sources pour les conditions de reproduction. AUCUNE garantie n'est donnée; tant pour des raisons COMMERCIALES que pour RÉPONDRE À UN BESOIN PARTICULIER.

Écrits par Roland McGrath et Ulrich Drepper.

The container has GLIBC_2.24. My system has GLIBC_2.35. Not sure which GLIBC is causing this issue. It most likely appears to be from container but then not sure how it worked earlier.

singularity version is 3.9.0

I also checked the container image with Fedora 35 and Ubuntu 22.04 and they all give GLIBC_2.34 error whereas Ubuntu 18.04 works well. So is this an issue with GLIBC on host machine?

As far as I know, the newer GLIBC should have older GLIBC binaries.

Can someone point to a solution to this?

CharlesM

Hello, have you found a solution ?

I am experiencing the same issue. (Glibc_ 2.34) Crashing in GPU, rendering problems in software rendering.

(For example, when it works, I cannot edit in text mode in the aster-study module. )

dbpatankar

CharlesM wrote:
Hello, have you found a solution ?

Not yet! Forced to stick to Ubuntu 18.04 for now until this gets resolved. Although I have no idea how thats going to happen because I was using it without any issue until a few weeks ago.

amit_code_aster

Hi dbpatankar and CharlesM

I got the same error on Fedora 36.

Another user @aurelius_nero also faced the same issue on Fedora 35. He solved it by installing nvidia-container-toolkit from NVIDIA website.

I also tried this solution and it worked for me also. The link to @aurelius_nero's post on this forum is https__://code-aster.org/forum2/viewtopic.php?id=26213 (remove __ as links are not allowed).

You can install `nvidia_container-toolkit` from the official NVIDIA repository https//nvidia.github.io/libnvidia-container/ (remove __ as links are not allowed). The repository also has instructions for how to install on Ubuntu 22.04.

Regards,

amit_code_aster

osman

I have ubuntu 22.04, nvidia rtx-3070Ti and the installation of nvidia-container-toolkit solved the problem of salome-meca only working with --soft option. Now, it works without --soft. Thanks for the info. This should be mentioned in the FAQ.

mpex

I'm experiencing a similar issue, though I don't have an NVIDIA GPU so I suspect the previously mentioned solutions will not work. Any suggestions?

*****************************************************

INFO : Running salome_meca in software rendering mode

*****************************************************

runSalome running on laptop

Searching for a free port for naming service: 2810 - OK

Searching Naming Service + found in 0.1 seconds

Searching /Kernel/Session in Naming Service +++SALOME_Session_Server: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.34' not found (required by /.singularity.d/libs/libGLX.so.0)

Traceback (most recent call last):

File "/opt/salome_meca/Salome-V2021-s9/modules/KERNEL_V9_7_0/bin/salome/orbmodule.py", line 181, in waitNSPID

os.kill(thePID,0)

ProcessLookupError: [Errno 3] No such process

During handling of the above exception, another exception occurred:

Traceback (most recent call last):

File "/opt/salome_meca/appli_V2021/bin/salome/runSalome.py", line 694, in useSalome

clt = startSalome(args, modules_list, modules_root_dir)

File "/opt/salome_meca/appli_V2021/bin/salome/runSalome.py", line 639, in startSalome

session=clt.waitNSPID("/Kernel/Session",mySessionServ.PID,SALOME.Session)

File "/opt/salome_meca/Salome-V2021-s9/modules/KERNEL_V9_7_0/bin/salome/orbmodule.py", line 183, in waitNSPID

raise RuntimeError("Process %d for %s not found" % (thePID,theName))

RuntimeError: Process 25480 for /Kernel/Session not found

--- Error during Salome launch ---

Humberto

Using Arch Linux

Dual GPU Inel + Nvidia

Same issue.

Works with --soft

**************************************


INFO : Running salome_meca in GPU mode
**************************************
runSalome running on archlinux
Searching for a free port for naming service: 2819 - OK
Searching Naming Service  + found in 0.1 seconds 
Searching /Kernel/Session in Naming Service  ++SALOME_Session_Server: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.34' not found (required by /.singularity.d/libs/libGLX.so.0)
Traceback (most recent call last):
  File "/opt/salome_meca/Salome-V2021-s9/modules/KERNEL_V9_7_0/bin/salome/orbmodule.py", line 181, in waitNSPID
    os.kill(thePID,0)
ProcessLookupError: [Errno 3] No such process

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/opt/salome_meca/appli_V2021/bin/salome/runSalome.py", line 694, in useSalome
    clt = startSalome(args, modules_list, modules_root_dir)
  File "/opt/salome_meca/appli_V2021/bin/salome/runSalome.py", line 639, in startSalome
    session=clt.waitNSPID("/Kernel/Session",mySessionServ.PID,SALOME.Session)
  File "/opt/salome_meca/Salome-V2021-s9/modules/KERNEL_V9_7_0/bin/salome/orbmodule.py", line 183, in waitNSPID
    raise RuntimeError("Process %d for %s not found" % (thePID,theName))

RuntimeError: Process 137243 for /Kernel/Session not found

amit_code_aster

Hi @Humberto.

A solution that worked for me on Fedora 36 is

1. Search for a file named `nvliblist.conf` in your installation. It should be under your Singularity installation directory somewhere.

2. Make a back-up of this file `mv nvliblist.conf nvliblist.conf.bak`.

3. Open the file `nvliblist.conf` using a text editor.

4. Delete the lines `libGLX.so.0`, `libglx.so.0`, and `libGLdispatch.so`.

Try running the Salome container again, it should work this time.

Reference: https: //github.com/apptainer/apptainer/issues/598

mf

Hello,

I also tried this, but: the mentioned lines ending with xxxx.so.0 do not exist in my nvliblist.conf in Ubuntu 22.04. All entries end with .so, it even says in the file, that this is mandatory.

So this did not solve the problem,

Mario.

trilbytim

This solution worked for me. After I installed the nvidia-container-toolkit it still didn't work. I didn't have the ".0" version lines listed in my file, so the lines I deleted from the "nvliblist.conf" file were: "libGLX.so", "libglx.so", and "libGLdispatch.so"

Also note that to make the backup copy use the command "cp nvliblist.conf nvliblist.conf.bak" (mv = move, so doesn't create a copy, in this case it moves it to the same place, i.e. just renames it

Hooray though, it does finally work for me, but this was a massive hassle to install, with all loads of dependencies to sort out before I even got singularity to work and then when that worked it had this problem!

Brigity

sudo nano /etc/singularity/nvliblist.conf to comment out libGLX.so as shown below worked for me on Ubuntu 22.04:

....

libglx.so

#libGLX.so

libnvcuvid.so

....

rmoortgat

Same problem here.

Manjaro 23.0.0

Nvidia container installed.

Tried all the here mentioned solutions without success.

It seems a bit odd to me that when GLIBC is updated to a newer version the program is broken.

But I'm no Linux specialist.

/usr/bin/nvidia-smi

**************************************

INFO : Running salome_meca in GPU mode

**************************************

runSalome running on roger-hpzbook17

Searching for a free port for naming service: 2811 - OK

Searching Naming Service + found in 0.1 seconds

Searching /Kernel/Session in Naming Service +++SALOME_Session_Server: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.34' not found (required by /.singularity.d/libs/libGLX.so.0)

Traceback (most recent call last):

File "/opt/salome_meca/V2022.1.0_scibian_univ/modules/KERNEL_V9_8_0/bin/salome/orbmodule.py", line 181, in waitNSPID

os.kill(thePID,0)

ProcessLookupError: [Errno 3] No such process

During handling of the above exception, another exception occurred:

Traceback (most recent call last):

File "/opt/salome_meca/appli_V2022.1.0_scibian_univ/bin/salome/runSalome.py", line 694, in useSalome

clt = startSalome(args, modules_list, modules_root_dir)

File "/opt/salome_meca/appli_V2022.1.0_scibian_univ/bin/salome/runSalome.py", line 639, in startSalome

session=clt.waitNSPID("/Kernel/Session",mySessionServ.PID,SALOME.Session)

File "/opt/salome_meca/V2022.1.0_scibian_univ/modules/KERNEL_V9_8_0/bin/salome/orbmodule.py", line 183, in waitNSPID

raise RuntimeError("Process %d for %s not found" % (thePID,theName))

RuntimeError: Process 5970 for /Kernel/Session not found

Thanks in advance,

Roger

hberro

Here is a fix for ubuntu 22.04 (and probably for other -recent- linux systems).

The problem is due to a (too recent) version of GLIBC required when the host system shares all of the libGL* files with the container.

The script in the following link must be placed in the folder where the .sif file and the launcher were created.

It extracts (through embedded tar data) libGL* libraries corresponding to the container OS (Scibian 9) (./sc9gl folder); it then runs salome_meca using the official launcher and by binding the sc9gl folder to overload the default singularity settings.

The link to download is :

[color=blue]drive.google.com/file/d/1ut8Loqn76u2Bqs6zVyrBHVsGqFNNjRxO[/color]

All feedback is welcome.

If you find this useful, we can upload it on code-aster.org

katoulos

hberro

Reporting success with salome_meca-lgpl-2022.1.0-1-20221225-scibian-9 on a SuSE Tumbleweed laptop

many thanks

mimis@hyugens:/opt/Salome-Meca $ inxi -G Graphics: Device-1: Intel CometLake-S GT2 [UHD Graphics 630] driver: i915 v: kernel Device-2: NVIDIA TU104M [GeForce RTX 2080 SUPER Mobile / Max-Q] driver: nvidia v: 550.78 Device-3: Realtek Integrated_Webcam_HD driver: uvcvideo type: USB Device-4: Tobii AB EyeChip driver: N/A type: USB Display: wayland server: X.org v: 1.21.1.12 with: Xwayland v: 24.1.0 compositor: kwin_wayland driver: X: loaded: nvidia unloaded: fbdev,modesetting,vesa dri: iris gpu: i915,nvidia resolution: 1: 3840x2160 2: 1920x1080 API: EGL v: 1.5 drivers: iris,nvidia,swrast,zink platforms: wayland,x11,surfaceless,device API: OpenGL v: 4.6.0 compat-v: 4.5 vendor: nvidia mesa v: 550.78 renderer: NVIDIA GeForce RTX 2080 Super/PCIe/SSE2 API: Vulkan v: 1.3.283 drivers: N/A surfaces: xcb,xlib,wayland

Kepsilone

hello,

perfect !!! thank you, it works very well.

best regards

rodrigofarias

hberro wrote:
Here is a fix for ubuntu 22.04 (and probably for other -recent- linux systems).

The problem is due to a (too recent) version of GLIBC required when the host system shares all of the libGL* files with the container.

The script in the following link must be placed in the folder where the .sif file and the launcher were created.

It extracts (through embedded tar data) libGL* libraries corresponding to the container OS (Scibian 9) (./sc9gl folder); it then runs salome_meca using the official launcher and by binding the sc9gl folder to overload the default singularity settings.

The link to download is :

[color=blue]drive.google.com/file/d/1ut8Loqn76u2Bqs6zVyrBHVsGqFNNjRxO[/color]

All feedback is welcome.

If you find this useful, we can upload it on code-aster.org

HB

Hi! Ive tried your script and came to the following error. What you think could be the problem?

@DESKTOP-MG4FINB:~$ sudo ./sm2022_libgl_patch_ub22.04.run . .. sc9gl.tgz sc9gl/ sc9gl/libGLX_mesa.so.0.0.0 sc9gl/libGLEW.so sc9gl/libGLdispatch.so sc9gl/libGLU.a sc9gl/libGLX_indirect.so.0 sc9gl/libGLdispatch.so.0.0.0 sc9gl/libGLESv2.so.2 sc9gl/libGLU.so.1 sc9gl/libGLEW.so.2.0 sc9gl/libGLU.so sc9gl/libGLX_mesa.so.0 sc9gl/libGL.so.1 sc9gl/libGLX.so.0.0.0 sc9gl/libGLESv1_CM.so.1.2.0 sc9gl/libGLESv2.so sc9gl/libGLdispatch.so.0 sc9gl/libGLEW.so.2.0.0 sc9gl/libGLX.so.0 sc9gl/libGLESv2.so.2.1.0 sc9gl/libGL.so.1.7.0 sc9gl/libGLESv1_CM.so sc9gl/libGLU.so.1.3.1 sc9gl/libGLX.so sc9gl/libGL.so sc9gl/libGLESv1_CM.so.1 Salome Launcher : ./.config/aster/salome_meca-lgpl-2021.json Running : ./.config/aster/salome_meca-lgpl-2021.json -B /home/rodrigo/sc9gl:/opt/prerequisites/gl/dispatch

./sm2022_libgl_patch_ub22.04.run: line 52: ./.config/aster/salome_meca-lgpl-2021.json: Permission denied

hberro

Hi rodrigofarias,

You should not run the script as root (or using sudo)... Just do :

./sm2022_libgl_patch_ub22.04.run

In the same folder as the sif and the launcher files.

If you have precedently run the command as sudo, then you might have to cleanup the directory sc9gl, as sudo using:

sudo rm -r sc9gl

Best regards,

benico

hberro

Thanks !

bash ./sm2022_libgl_patch_ub22.04.run
./salome_meca-20221225-scibian-9 -B "./sc9gl:/opt/prerequisites/gl/dispatch"

osman

Hello hberro,

Your solution works and very useful for linux installs. I had figured a solution in the past but have forgotten/erased my code-aster files. Now I was searching for a solution again, and found your scripts. You shopuld also add that the user should make it executable and run it on the host system.

Thanks again,

Osman

sb1966

Dear hberro,

Thanks a lot for your solution.

It has worked perfectly with Linux Mint 21.2 (code-named Victoria) in an old PC having NVIDIA GM108M [GeForce 920MX] card.

Best regards.

Sukumar

NorbertS

To run on Debian 11 and Debian 12 it is necessary to comment out libGLX.so in /etc/singularity/nvliblist.conf

libGLX_nvidia.so

libglx.so

#libGLX.so

libnvcuvid.so

Than it is possible to run salome-meca in GPU mode.

There is a singularity-container for Debian 11 and 12.

Add 'deb https: //fasttrack.debian.net/debian-fasttrack/ bullseye-fasttrack main contrib'

to /etc/apt/sources.list

Make 'apt install fasttrack-archive-keyring' before installing the singularity-container.

Norbert

scimulate

This isn't exactly a solution, but I'm running Salome-Meca through WSL on Windows 11 and ran into this issue when attempting to install on Ubuntu-22.04 and OpenSUSE Tumbleweed. For those doing similarly, install Oracle Linux 9.1 (glibc is version 2.34 on this), enable EPEL, install Singularity-CE, and then install the Salome-Meca Singularity container. That worked out-of-the-box for me.