Hi, On Thu, 12 Nov 2020 at 21:41, Ludovic Courtès <l...@gnu.org> wrote:
>> Hum? I do not know if I am doing correctly. The packages >> fakechroot-2.9-24.5.el6_1.1.x86_64.rpm and >> fakechroot-libs-2.9-24.5.el6_1.1.x86_64.rpm are installed. And I get >> as regular user: > > You do not need these packages: the tarball includes a copy of > libfakechroot.so (see > <https://hpc.guix.info/blog/2020/05/faster-relocatable-packs-with-fakechroot/>). > >> $ export GUIX_EXECUTION_ENGINE=fakechroot >> $ strace -f -s 500 -o logg ./bin/R >> fakechroot: unsupported Guix execution engine; ignoring > > You would need to use ‘guix pack -RR’ instead of ‘guix pack -R’ to get > the ‘fakechroot’ execution engine. > >> However, as root, simply running ./bin/R returns: >> >> # ./bin/R >> R: run.c:245: disallow_setgroups: Unexpected error: No such file or >> directory. >> Abandon > > That indicates that user namespaces are not supported. I have used ’-RR’ but because it is not working, then I tried to install extra stuff in case. But if the kernel is not supporting the namespace, that’s a end. Period. > But wait, if you’re root, you can just as well upgrade to a kernel that > supports unprivileged user namespaces… or even install Guix? :-) That’s not so simple. It is an old RHEL and I am not sure that the upgrading path is smooth… So I am applying the well-known rule: do not fix it if it is not broken. ;-) Well, I do not want to spend hours to figure out what is wrong with the new configuration (hardware). Installing Guix (the package manager) cannot work because the same problem: kernel. I am configuring another machine (Dell T620) to smooth the switch. But I hit the very same problem: spending hours to figure out what’s wrong because I am missing obscure hardware configurations. Thank you for your help. Cheers, simon
