Hello,
I have a number of applications that require GLIBC_2.18,
running them results in an error:
>/lib64/libc.so.6: version `GLIBC_2.18' not found
Finding the specific gclib version, shows 2.18 is not installed on the compute nodes.
>strings /usr/lib64/libc.so.6 | grep ^GLIBC_
...snip
GLIBC_2.14
GLIBC_2.15
GLIBC_2.16
GLIBC_2.17
GLIBC_PRIVATE
GLIBC_2.8
GLIBC_2.5
GLIBC_2.9
...snip
I know I can download and compile it on a normal machine, like:
>tar xf glibc-2.18.tar.gz
>cd glibc-2.18
>mkdir build && cd build
>../configure --prefix=/usr
>make -j2
>sudo make install
But it takes a long time to compile, is quite large, and is not
distributed to all the compute nodes. I could also do: yum update "glibc*"
but this requires root and I want to distribute the glibc-2.18 version
to all compute nodes at once, before I start using the gcp-slurm cluster.
Is there a way to use a different image, or should I include that into
a startup command, so I can use it before starting the cluster?
Thanks
Tobias