Hi,

I am trying to setup RStudio with conda locally in my WSL2 and seeing the following issue.

./start_rstudio_server.sh 8787
## Current env is >>
/home/pr/.miniconda3/envs/R
TTY detected. Printing informational message about logging configuration. Logging configuration loaded from '/etc/rstudio/logging.conf'. Logging to '/home/pr/.local/share/rstudio/log/rserver.log'.
2022-03-08T16:27:51.234866Z [rserver] ERROR system error 13 (Permission denied) [path: /var/run/rstudio-server, target-dir: ]; OCCURRED AT rstudio::core::Error rstudio::core::FilePath::createDirectory(const string&) const src/cpp/shared_core/FilePath.cpp:842; LOGGED FROM: int main(int, char* const*) src/cpp/server/ServerMain.cpp:626

The conda environment is showing up here correctly but the /var/run/rstudio-server doesn't exist.

I have modified the permission for the sqlite database as suggested in other issues.

ll /var/lib/rstudio-server
total 56
drwxr-xr-x  5 rstudio-server rstudio-server  4096 Mar  4 16:44 ./
drwxr-xr-x 36 root           root            4096 Mar  2 16:40 ../
drwxr-xr-x  2 root           root            4096 Mar  2 16:40 body/
drwxr-xr-x  2 root           root            4096 Mar  2 16:40 conf/
drwxr-xr-x  2 root           root            4096 Mar  2 16:40 proxy/
-rwxrwxrwx  1 rstudio-server rstudio-server 32768 Mar  4 16:44 rstudio-os.sqlite*
-rw-------  1 root           root              32 Mar  2 16:40 secure-cookie-key

I was able to get it to work with the local/start_rstudio_server.sh once over my home network but the R library and the binary was not picking up from conda even when the environment was active. However, the server now doesn't start, so I can't reproduce it.

Let me know if you need any additional information.

Thanks!

Hi,

I followed your nice script to bypass limitations of free rstudio server.
Actually I cannot manage to run it properly,

./start_rstudio_server.sh 8008

Current env is >>

/home/pkunderfranco/.conda/envs/Scenic_1.1.2.2

once I open in the web brower port 8008, R is not responding and keep loading,

Any suggestion?

Best
Paolo

Hi,
I'm very new to docker but I already find your repo greatly helpful.
I'm using a HPC system that I do not have root privilege, so I installed Singularity using conda.
Now I can run rserver in the login node. But what I really want to do is to be able to start a SLURM job and run RStudio server there.
When I tried to do this, I got the error:

FATAL:   while extracting rstudio_4.0.5.sif: root filesystem extraction failed: extract command failed: ERROR  : Failed to create user namespace: user namespace disabled
: exit status 1

I have no idea what this means. I tried the solution from rocker project page, and it's the same.
Therefore I'm wondering how should I run RStudio server in a computational node?

Thanks so much!

step to reproduce

1. clone this repo
2. run PORT=11231 PASSWORD=112233 ./run_singularity.sh
3. go to http://localhost:11231
4. enter username and password

result

expected result

enter the rstudio

more information

I can enter docker started pure rocker/rstudio
I've tested this repo on my own arch installation and a centos 7 server, same result

tried rocker/rstudio:3.6.3 problem exists

Bit of a newbie here, but when I run R studio server the regular way using r-base on ubuntu i can start and stop the server from terminal, also I can still access my terminal while running r studio server. But when I run it this way the terminal is not accessible and if I stop the session from R studio server menu nothing happens in command line?

I'm sorry for my foolish question.
I run the command line:

./start_rstudio_server.sh 8788

and it appear:

## Current env is >>
/home/XXX/anaconda3/envs/test_R

What should I do? I connect http://localhost:8788 and it still no connection. Thank you!

It would be nice to run rstudio server from a container:

• it's easier to setup if you don't have rstudio server already installed
• it potentially allows to run multiple instances even without rstudio server pro.

When running rstudio in podman, in had the issue that I could not access project directories on our scientific storage to
which I have access based on my secondary user group only.

This is considered a "security feature" of podman. A solution is described in this articel by RedHat. In brief

• use the crun instead of the runc runtime (install crun and edit ~/.config/containers/containers.conf)
• run the container with podman run --annotation run.oci.keep_original_groups=1 [...]

This appeared to work in principle (from a shell in the podman container I could access the group directory as expected),
but the rstudio process still got started with different permissions and didn't have access to the directory. I haven't yet figured out how to solve it (except by using Singularity instead (#13)).

Further issues:

• Due to a bug in podman 3.0.x, annotations are not read from containers.conf, making it impossible to pass the annotation to a container while using podman-compose. This is allegedly fixed in 3.1, but that version is not available on Centos 7.

Hi,

Thanks for writing such a foundational useful package - I've pretty much only used R through this package!

This is only tangentially related to this package but could be useful for others.

One difficulty I found was some lab members never wanted to switch environments and preferred to have rstudio auto-start as normal, but with the (base/other) conda environment already setup. It was too easy to forget to activate the conda environment and get subtly different versions.

And I realized I could use the same rsession tricks but at the systemctl level (if the system uses systemd). So

Modify /lib/systemd/system/rstudio-server.service:

ExecStart=/usr/lib/rstudio-server/bin/rserver
-> ExecStart=/root/activateConda

(Run systemctl daemon-reload when first change)

and
Create /root/activateConda: (made executable)

#!/bin/bash
eval "$(/home/jlanglie/anaconda3/bin/conda shell.bash hook)"
source /home/jlanglie/anaconda3/etc/profile.d/conda.sh

conda activate base
/usr/lib/rstudio-server/bin/rserver \
  --rsession-which-r=$(which R)     \
  --rsession-ld-library-path=$CONDA_PREFIX/lib

The main disadvantage is unless you modify that file, you can't easily change the conda environment. But if you did modify the file then you only have to do sudo systemctl restart rstudio-server.service for it to reload.

Thank you for the tool, it helped me a lot.
However,after I upgrade the rstudio-server, it seems there are some errors.

[rserver] ERROR Attempt to run server as user 'rstudio-server' (uid 999) from account 'aaa' (uid 1006) without privilege, which is required to run as a different uid; LOGGED FROM: virtual rstudio::core::ProgramStatus rstudio::server::Options::read(int, char* const*, std::ostream&) src/cpp/server/ServerOptions.cpp:322

Looking forward to your reply.

ubuntu@VM-8-7-ubuntu:~$ git clone [email protected]:grst/rstudio-server-conda.git
Cloning into 'rstudio-server-conda'...
[email protected]: Permission denied (publickey).
fatal: Could not read from remote repository.

Please make sure you have the correct access rights
and the repository exists.

Hi Gregor,

Many thanks for creating this repo. It is very useful for me.

Previously I used this repo (run locally) on a Centos 7.8 system, and everything is perfectly fine. But when I killed the service today and tried to use another port, I met the following error:

[root@localhost ~]# conda activate r4.1
(r4.1) [root@localhost local]# ./start_rstudio_server.sh 8787
## Current env is >>
/data1/Software/Installed/Anaconda3/envs/r4.1
TTY detected. Printing informational message about logging configuration. Logging configuration loaded from '/etc/rstudio/logging.conf'. Logging to '/var/log/rstudio/rstudio-server/rserver.log'.
2022-07-21T19:49:00.690672Z [rserver] ERROR system error 13 (Permission denied) [child-path: /data1/Software/PkgDownloads/rstudio-server-conda-master/local/rsession.sh]; OCCURRED AT rstudio::core::Error rstudio::core::system::runProcess(const string&, const string&, rstudio::core::system::ProcessConfig&, rstudio::core::system::ProcessConfigFilter) src/cpp/core/system/PosixSystem.cpp:2320; LOGGED FROM: rstudio::core::Error rstudio::core::system::launchChildProcess(std::string, std::string, rstudio::core::system::ProcessConfig, rstudio::core::system::ProcessConfigFilter, PidType*) src/cpp/core/system/PosixSystem.cpp:2168

The RStudio Server can still be accessed via web page, but once I log into my account, my terminal printed the above "ERROR system error 13" error, and I cannot open the R session. See the figure below:

Any suggestion or feedback would be appreciated!

I'm running your local scrip on WSL2 ubuntu 20.04 LTS. I first needed to set global write permissions on /run which then allows the script to run. Unfortunately it looks like the RETICULARE_PYTHON environmental variable is not being correctly set (I guess in rsession.sh) because rstudio server reports the error:

Error in py_discover_config(required_module, use_environment) : 
  Python specified in RETICULATE_PYTHON (/bin/python) does not exist

I also got a
CommandNotFoundError: Your shell has not been properly configured to use 'conda activate'.

error the first time I ran it. I thought that this might be the problem and that the conda env might not be being activated, but rstudio-server is finding the correct R. I ran conda init again anyway and it reported no changes to anything but I am also not getting that error any more so maybe thats sorted? Still getting the reticulate error though. Any ideas?

Hi @grst ,

thanks for the very useful repository!

I was just wondering if it would be possible to run the rstudio server within a docker container, providing the conda environment with all the dependencies within that container as well, so it can be ported to different instances and the analysis reproduced easily.

I've tried like this:

Dockerfile:

FROM rocker/rstudio

ENV LANG=C.UTF-8 LC_ALL=C.UTF-8
ENV CONDA_DIR /home/rstudio/conda

RUN apt-get update --fix-missing \
  && apt-get install -y wget bzip2 ca-certificates libglib2.0-0 libxext6 libsm6 libxrender1 git mercurial subversion \
  && apt-get clean

RUN wget \
    https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O /home/rstudio/miniconda.sh \
    && bash /home/rstudio/miniconda.sh -b -p /home/rstudio/conda \
    && rm -f /home/rstudio/miniconda.sh

ENV PATH=$CONDA_DIR/bin:$PATH
RUN which conda
RUN conda --version

COPY environment.yml /
RUN conda env create -f /environment.yml && conda clean -a -y

RUN conda env list
ENV PATH=$CONDA_DIR/envs/<envname>/bin:$PATH

ADD init2.sh /init2.sh

CMD ["/init2.sh"]

And docker-compose.yml:

version: "3.5"
services:
  rstudio:
    build: .
    # add the image name of your container
    image: <image-tag>
    ulimits:
      nofile: 10000
    ports:
        - "8889:8787"
    volumes:
      # mount the working directory containing your R project.
      - /mnt/volume/qsvac:/home/rstudio
      # share settings between rstudio instances
      - /home/centos/.local/share/rstudio/monitored/user-settings:/home/rstudio/.local/share/rstudio/monitored/user-settings
      # mount tempdir
      - /mnt/volume/rstudio-tmp/var/lib:/var/lib/rstudio-server
      - /mnt/volume/rstudio-tmp/var/run:/var/run/rstudio-server
      - /mnt/volume/rstudio-tmp/tmp:/tmp
    environment:
      - PASSWORD=notsafe
      # repeat the path of the conda environment
      - CONDAENV=/home/rstudio/conda/envs/<conda-env>

Also tested without sharing the rstudio settings and without mounting any tempdirs here. Unfortunately with no luck. If you happen to have an example on how to do that, that would be very helpful!

I have installed rstudio-server on my Ubuntu 20.04 workstation few hours ago. I could run sessions locally on the default port 8787. So far so good.
While attempting to run the server from a conda environment I encountered your solution on SO so I closed all sessions, stopped the server, and followed the recommended procedure.
Could you please help me understanding the returned error? See the screenshot below.

I thank you in advance for your kind feedback.

Thanks for your repository! It helps me a lot in the past two years. I'm using the singularity part

However, after I update R 4.1 to 4.2, the old version of rserver cannot draw figure in the plots channel anymore, and I upgraded to the latest version. I did run the singularity:

But I cannot login in the server on explorer, I've tried many combinations: root, whoami, rstudio, and their passwords, none of them success.

Do you have any idea? Thanks!

I get error:

./start_rstudio_server.sh: line 14: uuid: command not found
## Current env is >>
/mnt/anaconda3/envs/r_env
unrecognised option '--secure-cookie-key-file'

latest version, trying to run local version of /start_rstudio_server.sh 7070

rsession does not recognize conda environment packages

This is awesome and exactly what I need!
However, when running RStudio Server as a non root user, it seems to bypass the login page.
This is a big security risk for multi users systems.
Is there a way to have authentication functional by still running RStudio Server as a non root user?

Can we, as an alternative to conda, also specify a docker/singularity container with R and some libraries preinstalled?

With the latest version of RStudio:

wget https://download2.rstudio.org/server/centos6/x86_64/rstudio-server-rhel-1.2.5033-x86_64.rpm
sudo yum install rstudio-server-rhel-1.2.5033-x86_64.rpm

I am getting the following error when trying to launch RStudio Server:

error while loading shared libraries: libssl.so.1.1: cannot open shared object file: No such file or directory

Any Solution? It might be due to a " libssl.so.1.1" installation issue probably.

Hi,

I was interested in using this but got consistent errors that CommandNotFoundError: Your shell has not been properly configured to use 'conda activate'., tracing back to the conda activate line in rsession.sh. After some debugging and following conda/conda#7980, I found replacing
conda activate ${CONDA_ENV}
with
source ~/anaconda3/bin/activate ${CONDA_ENV}
fixed it for me. I'm happy to write up a small PR. This issue might help someone else in case they also have this problem.

When connecting to some AWS's EMR Clusters, I am getting "secure-cookie-key-file" error, as not being a recognized launch statement.

Is this a know behaviour?

By skipping such --secure-cookie-key-file option I am able to launch RStudio Server however. Is it safe to do so?

Thanks.
Juan

Hi,
I have used your scripts to run Rstudio server in my docker container setup, thank you so much! But with the new 1.4 release of RStudio, they changed something when it comes to the sqlite database. I don't know why this only becomes a problem with the newest release.

After the update, Rstudio server just exited after starting and reports:

"ERROR database error 7 (sqlite3_statement_backend::loadOne: attempt to write a readonly database) [description: Could not delete expired revoked cookies from the database, description: Could not read revoked cookies from the database]; OCCURRED AT virtual rstudio::core::Error rstudio::core::database::Connection::execute(rstudio::core::database::Query&, bool*) src/cpp/core/Database.cpp:480; LOGGED FROM: int main(int, char* const*) src/cpp/server/ServerMain.cpp:729

This seems to be related to https://community.rstudio.com/t/permissions-related-to-upgrade-to-rstudio-server-open-source-1-4/94256/3 and rocker-org/rocker-versioned2#105

The directory /var/lib/rstudio-server contains the sqlite file (/var/lib/rstudio-server/rstudio.sqlite) which needs to be writeable by the user running the server and has a default system daemon user GID UID set by the installer (at least in the deb file).

I fixed it by giving the file to my user executing rstudio server, but deleting the file and allowing other users to write in the parent directory apparently also works.

The better fix may be to give rstudio server the --database-config-file arg flag and provide a config file like this

directory=/var/lib/rstudio-server 

In the default file, all options are commented, so it is basically empty.

Hopefully this helps someone,
Lukas Jansen