--- title: Important Changes for HPCC Users linkTitle: Important Changes for HPCC Users type: docs aliases: - /changes.html - /changes --- ## Rollout of Rocky and DUO: Feb 18 through Mar 17, 2022
Also see email notification sent to users on 18-Feb-2022.


This is to inform users about several important system upgrades the HPCC will be implementing in the next 30 days (starting February 18th, 2022). Importantly, these changes are relevant to all users including those who are accessing the HPCC systems via JupyterHub, RStudio Server, SSH or sFTP/SCP. Thus, please read the instructions below carefully. If anything is unclear or there are questions, please email support@hpcc.ucr.edu for help. The most important change will be the switch from the old `CentOS/RHEL 7` platform to the new `Rocky/RHEL 8`. We anticipate to finalize this upgrade on March 17th, 2022. This gives users a 30-day transition period to log into the new `Rocky/RHEL 8` platform and test whether the software and other resouces they are using for their research are properly working. It is important to understand the deployment of Rocky 8 is a major upgrade that requires the systems administrators recompiling most software from the old system onto the new system. To avoid unnecessary extra downtimes, we are also elevating with this upgrade our security standards by adopting UCR’s DUO multi factor authentication system. This is important to prevent intrusions and comply with UC-wide IT standards. ### Operating System As mentioned above, the biggest change is that we are upgrading the OS from `CentOS/RHEL 7` to `Rocky/RHEL 8`. [Rocky Linux](https://rockylinux.org/) is the community equivalent and identical to `RHEL` (similar to how `CentOS` was). Currently, `pigeon` is the only head/login node that is linked to the new `Rocky/RHEL 8` platform. To check which platform you are on, you can run the `platform` command. The upgrade from `RHEL 7` to `RHEL 8` will result in the following user-facing changes. ### Passwords User passwords will be expired on the current `CentOS/RHEL 7` platform and users will need to reset their password during the next login. During this login, users need to provide their old password twice. The first time is to authenticate as cluster user (login password), and the second time to authorize the password reset (kerberos password). After this users will be prompted to provide their new password . When logging into the new `Rocky/RHEL 8` platform as outlined [here](https://hpcc.ucr.edu/manuals/hpc_cluster/login/), users also need to configure `DUO` if using a password, or alternatively create an SSH key pair. If a user is new to DUO, the instructions from UCR's ITS are [here](https://bit.ly/3JFIKu9). Users accessing the cluster via SSH key pairs are strongly encouraged to also reset their password upon login by following the instructions [here](/manuals/hpc_cluster/start/#change-password). Additional information about login related topics are provided on this [manual page](/manuals/hpc_cluster/login/). ### External Users External users are unlikely to have a UCR NetID required for DUO. This includes users with HPCC cluster and restricted data transfer accounts. Those users want to access the cluster via SSH keys. This is both convenient (no need to type a password anymore) and secure. Please refer to our [SSH keys](/manuals/hpc_cluster/sshkeys/) manual for detailed instructions of configuring SSH key-based access. After creating an ssh key pair, users will need to email their public SSH key to support@hpcc.ucr.edu so that the systems admin can add their public SSH key to the corresponding user account. Note, the following instructions are only relevant for users who perform computations on our cluster(s). Users who are using our systems exclusively for data transfers can ignore them. ### Software All software installation requests will be restricted to the new cluster. New software may be installed under a different version, or may no longer be installed as a module, or may not be installed at all. Run the following command to list currently available modules: ```bash module avail ``` Or you can search for a specific software (ie. `vim`) like so: ```bash module avail vim ``` ```bash which vim ``` Since the new platform is built using `GCC 8.5`, then all previous compiled software must be re-compiled on the new platform. If you cannot find what you are looking for send an installation request to [support@hpcc.ucr.edu](mailto:support@hpcc.ucr.edu). Compatibility mode (`singularity`) can be used to run the older `CentOS 7` modules, however this may not work in all cases. Please refer to our [Singularity Examples](/manuals/hpc_cluster/singularity/) for more information. ### Bash We officially support bash, even though other shells may work they have not been tested under the new `Rocky/RHEL 8` platform. When logging in under a bash shell, some errors/warnings may be visible. The most common message being that a module cannot be loaded. Check if the module (ie. `vim`) is available with the following: ```bash module avail vim ``` If there is no output, then the module is not currently available. Either remove the `module load vim` from your `~/.bashrc` and/or `~/.bash_profile` files, or request that it be installed. If the software is available, ensure a proper check is in place around loading modules within your `~/.bashrc` and/or `~/.bash_profile` files. For example: ```bash if [[ "$(type -t module)" == "function" ]]; then module load tmux module load neovim fi ``` It may also help to keep `~/.bashrc` free of unnecessary bloat and only add customized changes to `~/.bash_profile`. Also keep in mind that when running jobs with just `/bin/bash` the `~/.bashrc` file is loaded. However, adding the lower case `L` to a job's interpreter, as in `/bin/bash -l` this will load the `~/.bash_profile` file. This can be useful since it provides flexibility to initialize a default job shell or a customized job shell. ### Slurm A newer version of Slurm is being used on the new `Rocky/RHEL 8` platform, however very little is different from the previous version. All previous job submission scripts and commands/flags should still be compatible. During the transition period From the old `CentOS/RHEL 7` platform any Slurm jobs scheduled to start after March 17th will never start. Please check your jobs and ensure that they run before this time frame. You can check your start times with the following command: ```bash squeue --start -u $USER ``` Be sure to move to the newer `Rocky/RHEL 8` platform as soon as possible. ## User-facing Changes Implemented on 23-Aug-2019 ### Domain Names The old domain names `biocluster.ucr.edu` and `bioinfo.ucr.edu` have finally been discontinued. As a result, users need to use the new `hpcc.ucr.edu` name for the following services: + _ssh/scp/ftp/http_: `cluster.hpcc.ucr.edu` instead of `biocluster.ucr.edu` + _RStudio Server_: `rstudio.hpcc.ucr.edu` instead of `rstudio.bioinfo.ucr.edu` + _Jupyter_: `jupyter.hpcc.ucr.edu` instead of `jupyter.bioinfo.ucr.edu` + _Email Support_: `support@hpcc.ucr.edu` instead of `support@biocluster.ucr.edu` In addition, URLs containing `biocluster.ucr.edu` need to be updated to `cluster.hpcc.ucr.edu`. ### Password Reset After the upgrade on Aug 23, 2019, all HPCC users have been emailed a temporary password with instructions how to change it. This email was sent to the address we have on file for each user. In case you missed the corresponding email notification and/or you are not able to log into the cluster, please email us at [support@hpcc.ucr.edu](mailto:support@hpcc.ucr.edu) to receive a new password. ### Updated OpenMPI If you have compiled or use software that was compiled with OpenMPI, then it will need to be recompiled. If you are running into any issues, please email us at [support@hpcc.ucr.edu](mailto:support@hpcc.ucr.edu).