Useful hints:
The most common problem with X programs is wrongly set DISPLAY variable.
Check your IP address (ifconfig for Linux, and ipconfig for Windows)
For bash:
export DISPLAY=computer_name:0 (instead of computer name you can us IP address)
For csh or tcsh:
setenv DISPLAY computer_name:0
How to install USB devices - memory stick, external USB disk, etc.
Insert memory stick and than check under which partition is located (usually /dev/sda1): fdisk -l /dev/sda
If you have already another SCSI device than it will be the following SCSI device sdb, sdc...
It is the best idea to format it with FAT32 than you can use it under Windows and Linux
Edit /etc/fstab and add following line:
/dev/sda1 /mnt/usb vfat default,users 0 0
Create mount point /mnt/usb (or whatever name you prefer). Than mount memory stick:
mount /mnt/usb
Check with df if your memory stick was properly mounted.
Before removing memory stick: umout /mnt/usb because if device is not unmounted you can loose your data.
More detailed guied for working with USB devices
Looking for big or old files and deleting them
If you want to decrease disk usage you can localize big files and delete them if not needed:
find /your_directory -size +10000k -exec ls -la {} \; (Finds all files bigger than 10 MB and displays them)
find /your_directory -size +10000k -exec rm -f {} \; (Finds all files bigger than 10MB and deletes them)
find /your_directory -atime +30 -exec rm -f {} \; (Finds all files not accessed during 30 days and deletes them)
After -exec statement you can put other commands like: gzip, ls or rm
For bash:
export TMP=/scratch/$USER (sets TMP variable as /scratch/your_user_name)
For csh or tcsh:
setenv TMP /scratch/$USER
Queuing system - submitting jobs and monitoring free resources
To submit jobs on the cluster type:
qsub /apps/bin/run_script job_name This command will submit program run_script with the job job_name
You can copy one of the scripts located in /apps/bin and modify it for your needs; remember to change path for your script
To submit job to specific node use
qsub -q node12.q /apps/bin/run_script job_name this command will submit job to node12
Killing submitted jobs
To kill the submitted job first type: qstat |sort
to obtain ID of the job and than type:
qdel ID_of_the_job
To monitor running jobs on the cluster type:
qstat display running jobs on the cluster (use qstat |sort to sort jobs with the submission time)
qstat -f |more display status for all nodes. Average load shows usage of specific node
qacct -o -d 14 queue accounting. Display usage of the cluster for all users in last 14 days
qacct -g -d 7 Display usage of the cluster for all groups in last 7 days
Automatic jobs - your own cron jobs
You can create your own automatic jobs which will execute periodically
Use crontab -e command and set the timing when your job should execute:
00 8 * * * * your_shell_script executes your_shell script daily at 8:00 AM
What is Quota and disk usage monitoring
Disk quota limits the user disk space usage. For standard user we set it up to: 1.3GB soft limit and 1.5 GB hard limit. When you exceed soft limit you have one week to clean up your files.
You will be automatically notify with the e-mail that you are exceeded the quota limits.
You won't be allowed to use more space than hard limit. If you have specific needs those limit can be increased by reasonable number.
To check your quota type: quota It displays your quota limit and your disk usage
To decrease your disk usage you can locate your big files and compress them with gzip command
How to find rpm's and solve problems with your program errors:
To find specific rpm try to locate it on the web (use google and search for your rpm and put system name too) or use one of the sites below:
http://rpm.pbone.net
http://www.rpmfind.net/linux/RPM/index.html
If you have a problem with your program I suggest to check the web first (probably somebody had the same problem before).
Use google and search for your error message (it usually give you solution to your problem).
Creating authorized keys for ssh - no more password typing
You can create your public key and use it to authorize yourself with all the nodes. Follow this procedure:
ssh-keygen -t rsa Do not put any phrase
Than copy public key (usually id_dsa.pub) to authorized_keys:
cat id_dsa.pub > authorized_keys
First time you ssh to the node you need to answer yes to add this node to authorized clients.
To make it automatic for all nodes use for instance Runall date or dsh -a date and answer yes for all nodes
To locate files you can use one of the following commands:
locate file
slocate file
find / -name name*
ssh, sftp, scp - secure protocols
ssh, sftp, scp are replacement of non secure protocols like: telnet, ftp. They are using crypted session (port 22) and data are transmitted in crypted form between two computers.
usage:
ssh and sftp are very similar to telnet and ftp protocols
ssh username@computer_name
sftp username@computer_name
scp file_name computer.domain.com:/directory/subdirectory This command will transfer file file_name to computer.domain.com with specified directory
scp -r directory_name computer.domain.com:/directory/subdirectory Corresponding syntax for directory transfer