Unable to load libjulia.so because `GLIBCXX_3.4.20' not found

[Non-Contradiction]

After library(JuliaCall) and julia_setup(), I encountered this error:

Error in dyn.load(.julia$dll_file, FALSE, TRUE) :
unable to load shared object '/storage/work/cxl508/julia/julia-0.6.0/bin/../lib/libjulia.so.0.6':
/usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.20' not found (required by /storage/work/cxl508/julia/julia-0.6.0/bin/../lib/libjulia.so.0.6)

And this is the sessionInfo:

sessionInfo()
R version 3.4.0 (2017-04-21)
Platform: x86_64-redhat-linux-gnu (64-bit)
Running under: Red Hat Enterprise Linux Server release 6.9 (Santiago)

Matrix products: default
BLAS: /usr/lib64/R/lib/libRblas.so
LAPACK: /usr/lib64/R/lib/libRlapack.so

locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

attached base packages:
[1] stats graphics grDevices utils datasets methods base

other attached packages:
[1] JuliaCall_0.7.3

loaded via a namespace (and not attached):
[1] compiler_3.4.0

[Non-Contradiction]

The issue happens because libstdc++.so.6 is too old, update libstdc++6 could solve this. Or maybe you could add the directory $JuliaDir/lib/julia to LD_LIBRARY_PATH which contains all the shared object libjulia needs, for example,
export LD_LIBRARY_PATH=$JuliaDir/lib/julia:$LD_LIBRARY_PATH

[Non-Contradiction]

Maybe we could add more helpful error messages in this part.

[matbesancon]

@Non-Contradiction this seems to be the closest issue to what I experience:

R aborting with:

> library(JuliaCall)
> julia_setup()
Julia version 1.1.0 at location /home/user/programs/julia/bin will be used.

  unable to load shared object '/home/user/programs/julia/bin/../lib/libjulia.so.1':
  /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version `GLIBCXX_3.4.22' not found (required by /home/user/programs/julia/bin/../lib/julia/libLLVM-6.0.so)

 *** caught segfault ***
address 0x38, cause 'memory not mapped'

For info:

> version
               _                           
platform       x86_64-pc-linux-gnu         
arch           x86_64                      
os             linux-gnu                   
system         x86_64, linux-gnu           
status                                     
major          3                           
minor          4.4                         
year           2018                        
month          03                          
day            15                          
svn rev        74408                       
language       R                           
version.string R version 3.4.4 (2018-03-15)
nickname       Someone to Lean On          

[Non-Contradiction]

Thank you very much for the feedback.
The problem is because R cannot find the version of libstdc++.so.6 that is needed by Julia.
The solution is to export the path /home/user/programs/julia/bin/../lib/ into LD_LIBRARY_PATH or R_LD_LIBRARY_PATH.
I thought I have added some error handling and messages for this... but the error handling doesn't work here. I will add a part of this in README in the troubleshooting section.

[Non-Contradiction]

The commit ebb3a46 has the issue and the solution in README.

[matbesancon]

One useful command for people to check if this works:
strings /usr/lib/x86_64-linux-gnu/libstdc++.so.6 | grep GLIBCXX, there should be GLIBCXX_3.4.22 in the suggestions

[s439]

When running julia_setup() through opencpu I am getting the error

"Error in juliacall_initialize(.julia$dll_file) : 
/usr/julia/julia-d386e40c17/bin/../lib/libjulia.so.0.6 - /lib64/libstdc++.so.6: version `GLIBCXX_3.4.20' not found (required by /usr/julia/julia-d386e40c17/bin/../lib/libjulia.so.0.6)"

I have set environment varibale R_LD_LIBRARY_PATH in /etc/bashrc and in /etc/opencpu/Renviron aswell
Am I missing something??
Please HELP!

[Non-Contradiction]

@s439 Thanks for the feedback.
From the error message, we can see that the error happens because libstdc++ in the system is still used instead of the libstdc++ library in the Julia folder.
What is the result of system("$R_LD_LIBRARY_PATH;") in the R session?
How about setting LD_LIBRARY_PATH?

[s439]

@Non-Contradiction The result is empty "" through opencpu session
I have set both the environment variables LD_LIBRARY_PATH & R_LD_LIBRARY_PATH in all different ways (Including setting the EVs in opencpu session). But no luck.
For the julia_setup() command, I am able to provide JULIA_HOME through the JULIA_HOME parameter.
In the same way, is it possible to forcefully point the libstdc++ library in the Julia folder? (through a parameter of julia_setup() command)

[Non-Contradiction]

Then it is quite strange, it seems that the R doesn't pick up the R_LD_LIBRARY_PATH, maybe we can try export R_LD_LIBRARY_PATH = blah blah blah and then start R and try julia_setup()?
Currently, it is not possible to set the library path for Julia, but I will add a parameter in the julia_setup() soon.

[mattcbro]

This issue appears to still be a problem here in 9/9/2020. Sys.getenv() in R, shows that R_LD_LIBRARY_PATH does not exist. Though LD_LIBRARY_PATH does; and the latter can be set with quite a bit of difficulty using Sys.setenv(). The difficulty lies in the fact that you want to do this programmatically using the current values of LD_LIBRARY_PATH.

Nevertheless when I add my local julia/lib directory to this, I get past the GLIBCXX error, but then I start running into a bunch of other missing symbols and libraries.

I now get this error:

Error: Error happens in Julia.
InitError: could not load library "/home/matt/.julia/artifacts/079e2f87254bb2465a44de255cd5dde50d4d0b62/lib/libpangocairo-1.0.so"
/usr/lib/x86_64-linux-gnu/libpangoft2-1.0.so.0: undefined symbol: pango_font_get_hb_font
Stacktrace:
 [1] dlopen(::String, ::UInt32; throw_error::Bool) at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.5/Libdl/src/Libdl.jl:109
 [2] dlopen at /buildworker/worker/package_linux64/build/usr/share/julia/stdlib/v1.5/Libdl/src/Libdl.jl:109 [inlined] (repeats 2 times)
 [3] __init__() at /home/matt/.julia/packages/Pango_jll/sDpKp/src/wrappers/x86_64-linux-gnu.jl:82
 [4] _include_from_serialized(::String, ::Array{Any,1}) at ./loading.jl:697
 [5] _require_search_from_serialized(::Base.PkgId, ::String) at ./loading.jl:782
 [6] _tryrequire_from_serialized(::Base.PkgId, ::UInt64, ::String) at ./loading.jl:712
 [7] _require_search_from_serialized(::Base.PkgId, ::String) at ./loading.jl:771
 [8] _tryrequire_from_serialize

My LD_LIBRARY_PATH does contain the path to the julia libraries, namely.

> Sys.getenv("LD_LIBRARY_PATH")
[1] "/home/matt/programs/julia-1.5.1/lib/julia:/usr/lib/R/lib::/lib:/usr/lib/x86_64-linux-gnu:/usr/lib/jvm/java-8-oracle/lib/server"

[Non-Contradiction]

@mattcbro Maybe we could try to add the julia library path after the library path of R? Or maybe the end of LD_LIBRARY_PATH?

[mattcbro]

The problems is rather subtle. I get library symbol failures whenever I try to use the Plots package inside a user defined module. Somehow the environment changes so that it can't find the libraries in that situation.

I will try to generate a simple example tomorrow.

[mattcbro]

OK I opened a new issue and provided example code.
#136

[biona001]

I had the same exact issue on a cluster which load Julia via the module load Julia command.

For future reference, first I typed which julia which gave the Julia executable path, and it in my .bash_profile as

export R_LD_LIBRARY_PATH=/share/software/user/open/julia/1.6.2/lib/julia

Then JuliaCall works

library(JuliaCall)
julia <- julia_setup(JULIA_HOME = "/share/software/user/open/julia/1.6.2/bin")
julia_exists("sqrt")

[1] TRUE