This information is for users to migrate the previous system to the new system replaced in fiscal 2022.
There are no changes from the previous system in the host name (round robin) and how to log in. If you want to log in to the login node directly without going through the round robin, please refer to Access because the specific host name has been changed.
If the following message appears and you cannot log in, you need to delete the known_hosts information.
@@@ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @@@
IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY!
Someone could be eavesdropping on you right now (man-in-the-middle attack)!
It is also possible that a host key has just been changed.
The fingerprint for the RSA key sent by the remote host is
c0:30:d6:93:b2:d8:06:4a:6f:9c:d5:00:cc:c5:69:58.
Please contact your system administrator.
Add correct host key in /home/xxx/.ssh/known_hosts to get rid of this
message.
Offending RSA key in /home/xxx/.ssh/known_hosts:3
RSA host key for laurel.kudpc.kyoto-u.ac.jp has changed and you have
requested strict checking.
Host key verification failed.You can delete the known_host information with the following instruction.
$ ssh-keygen -R laurel.kudpc.kyoto-u.ac.jp%homepath%\.ssh\known_hosts (Windows), /Users/(username)/.ssh/known_hosts (Mac, Linux) using an editor.%appdata%\MobaXterm\MobaXterm.ini in an editor.ssh-ed25519@22:laurel.kudpc.kyoto-u.ac.jp=0xd152edcd(omitting the following)The configuration of modules, location of applications and environment variables have changed since the previous system.
If you used a customized .bashrc in the previous system, please modify the .bashrc as necessary.
You can also initialize the .bashrc by copying /etc/skel/.bashrc to your home directory as follows.
If you cannot log in, please let us know using the Inquiries Form, and we will initialize the shell configuration file with administrator privileges.
$ cp /etc/skel/.bashrc $HOMEFrom the new system, we have changed the operation to manage SSH public keys all together in the user portal. Accordingly, the .ssh directory in the home directory ($HOME) has been moved under $HOME/DOTFILES_20221108/. If you do not need it, please delete it.
User data that had been saved in the previous system was migrated automatically.
/LARGE2 has been consolidated into /LARGE0 and /LARGE3 into /LARGE1. We configure that you can link to /LARGE0 from /LARGE2 and link to /LARGE1 from /LARGE3 with the existing path, however the configuration will be removed in the future, so please update your path.
In addition, quota management for large volume storage has been changed from Group Quota to Project Quota. As a result, capacity is managed in the unit of paths of the large volume storage group, not the group to which the file belongs.
For details on the file system configuration of the new system, please refer to Use of Storage.
CPU time and the amount of memory for each system's login node is extended in order to avoid interruptions during file transfers to PCs.
| System | CPU time (standard) | CPU time (maximum) | Amount of memory (standard) |
|---|---|---|---|
| Previous system | 4 hours | 24 hours | 8GB |
| New system | 4 hours | 24 hours | 16GB |
The OS will be changed from CLE/RHEL 7 to RHEL 8.
Compilers will be provided for Intel, NVIDIA HPC SDK, and GNU. The Cray compiler will no longer be provided.
The job scheduler will be changed from PBS to Slurm.
| Purpose | PBS | Slurm |
|---|---|---|
| Specify the queue to submit jobs | -q QUEUENAME | -p QUEUENAME |
| Specify the execution group | -ug GROUPNAME | Not required |
| Specify the elapsed time limit | -W HOUR : MIN | -t HOUR:MIN |
| ・ Specify the number of processes ・ Specify the number of threads per process ・ Specify the number of CPU cores per process ・ Specify the memory size per process |
-A p=X:t=X:c=X:m=X | --rsc p=X:t=X:c=X:m=X |
| Specify the standard output file name | -o FILENAME | Not changed |
| Specify standard error output file name | -e FILENAME | Not changed |
| Summarize standard error output | -j oe(Output to standard output) / eo(Output to standard error) | Not changed |
| Send Email | -m a(when a job interrupted) / b(When started) / e(When ended) | --mail-type=BEGIN(When started)/END(when ended)/FAIL(When a job interrupted)/REQUEUE(when re-executed)/ALL(All) |
| Specify email address | -M MAILADDR | --mail-user=MAILADDR |
| Specify prohibition of job re-execution when failure occurs | -r n | --no-requeue |
| Purpose | PBS | Slurm |
|---|---|---|
| Check the queue where jobs can be submitted | qstat -q | spartition |
| Submit a job to the queue | qsub | sbatch |
| Check job status | qstat | squeue |
| Cancel a submitted job. | qdel | scancel |
| Check job details | qs | sacct -l |
| Purpose | PBS | Slurm |
|---|---|---|
| Job ID | QSUB_JOBID | SLURM_JOBID |
| Name of the queue where the job was submitted | QSUB_QUEUE | SLURM_JOB_PARTITION |
| Current directory where the job was submitted | QSUB_WORKDIR | SLURM_SUBMIT_DIR |
| Number of processes allocated when executing a job | QSUB_PROCS | SLURM_DPC_NPROCS |
| Number of threads allocated per process when executing a job | QSUB_THREADS | SLURM_DPC_THREADS |
| Number of CPU cores allocated per process when executing a job | QSUB_CPUS | SLURM_DPC_CPUS |
| Upper limit for the amount of memory allocated per process when executing a job | QSUB_MEMORY | ー |
| Number of processes placed per node when executing a job | QSUB_PPN | ー |
You can convert job script commands and options used in the PBS environment for Slurm with the pbs2slurm command.
pbs2slurm input_script [output_script][b59999@camphor1 script]$ cat pbs.sh
#!/bin/bash
#======Option========
#QSUB -q gr19999b
#QSUB -A p=1:t=1:c=1:m=1G
#QSUB -W 12:00
#QSUB -r n
#QSUB -M kyodai.taro.1a@kyoto-u.ac.jp
#QSUB -m be
#====Shell Script====
mpiexec.hydra ./a.out
[b59999@camphor1 script]$ pbs2slurm pbs.sh slurm.sh
[b59999@camphor1 script]$ cat slurm.sh
#!/bin/bash
#======Option========
#SBATCH -p gr19999b
#SBATCH --rsc p=1:t=1:c=1:m=1G
#SBATCH -t 12:00
#SBATCH --no-requeue
#SBATCH --mail-user=kyodai.taro.1a@kyoto-u.ac.jp
#SBATCH --mail-type=BEGIN,END
#====Shell Script====
srun ./a.outThe pbs2slurm command supports conversion of the following options. Options not included below should be modified individually.
| Before conversion | After conversion | Purpose |
|---|---|---|
| #QSUB -q | #SBATCH -p | Specify queues |
| #QSUB -A | #SBATCH --rsc | Specify resources |
| #QSUB -W | #SBATCH -t | Specify elapsed time |
| #QSUB -N | #SBATCH -J | Specify job name |
| #QSUB -o | #SBATCH -o | Specify the destination for standard output |
| #QSUB -e | #SBATCH -e | Specify the destination for standard error output |
| #QSUB -m | #SBATCH --mail-type | Specify the timing of email sending |
| #QSUB -M | #SBATCH --mail-user | Specify the recipient of the email |
| #QSUB -r n | #SBATCH --no-requeue | prohibition of job re-execution |
| #QSUB -J | #SBATCH -a | Specify array job |
| mpiexec | srun | MPI Execution(If there is a option, it must be removed manually.) |
| mpiexec.hydra | srun | MPI Execution(If there is a option, it must be removed manually.) |