Permanently setting users PATH
The CCG Ubuntu server uses /home/user/.bashrc
With admin priveledges, edit .bashr lines:
#PATH info
export PATH=/opt/openmpi:$PATH
export PATH=/opt/:$PATH
export PATH=/opt/openmpi/bin:$PATH
export LD_LIBRARY_PATH=/opt/openmpi/lib:$LD_LIBRARY_PATH
export PATH=/opt/openmpi/etc:$PATH
Set up Galaxy server to autostart on boot.
Edit the /data/galaxy/galaxy-dist/contrib/galaxy.debian-init sample file to refelct the correct path to Galaxy /data/galaxy/galaxy-dist.
Copy the file to /etc/init.d/galaxy:
$ sudo cp /data/galaxy/galaxy-dist/contrib/galaxy.debian-init /etc/init.d/galaxy
Configure for start at boot with:& update-rc.d galaxy defaults
The screen should print:Adding system startup for /etc/init.d/galaxy ...
/etc/rc0.d/K20galaxy -> ../init.d/galaxy
/etc/rc1.d/K20galaxy -> ../init.d/galaxy
/etc/rc6.d/K20galaxy -> ../init.d/galaxy
/etc/rc2.d/S20galaxy -> ../init.d/galaxy
/etc/rc3.d/S20galaxy -> ../init.d/galaxy
/etc/rc4.d/S20galaxy -> ../init.d/galaxy
/etc/rc5.d/S20galaxy -> ../init.d/galaxy
installing Nvidia CUDA driver to enable use of BEAGLE libraries for apps using GPU processing.
I followed Utkarsh Jaiswal's excellent online instructions.
The first thing to do is to confirm which Nvidia drivers are installed:
$ lsmod | grep video
If the returns something like:$ Video 19596 nouvaeu
this means you have the deafault open source Nvidia drivers. If the command returns nothing, try:
$ lsmod | grep nvidia
if you see something similar to: Nvidia 11270980 38
this means you have the proprietary CUDA drivers installed. I had to install the requisite tools (not sure what these are):
$ sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx
libglu1-mesa libglu1-mesa-dev
I also had to "blacklist" required modules:
$ sudo vi /etc/modprobe.d/blacklist.conf
and add the following lines at the end of the file:
blacklist amd76x_edac
blacklist vga16fb
blacklist nouveau
blacklist rivafb
blacklist nvidiafb
blacklist rivatv
To clean up Nvidia files sudo apt-get remove --purge nvidia*
I then downloaded the LINUX: CUDA 5.0 Production Release version "cuda_5.0.35_linux_64_ubuntu11.10-1.run" for Ubuntu 11.10
(Our current Ubuntu version is 12.04) into ~/Downloads/.
I changed permissions of "cuda_5.0.35_linux_64_ubuntu11.10-1.run" $ cd ~/Downloads
$ sudo chmod +x cuda_5.0.35_linux_64_ubuntu11.10-1.run
Then reboot. Do not log into the GUI interface, enter Ctrl+Alt+F1 to use terminal login.
The CUDA drivers will not install if the display manager is running (ours is LightDM). So kill this before installinig. $ sudo service lightdm stop
Now run the cuda file "cuda_5.0.35_linux_64_ubuntu11.10-1.run"
$ sudo cuda_5.0.35_linux_64_ubuntu11.10-1.run
You will be presented with the option to accept a license agreement. And you will be presented with three option of installing the
drivers, the toolkit and the samples. install all but samples, we will not need them.
Now edit the ~/.bashrc file and add:export LD_LIBRARY_PATH=/usr/local/cuda/-5.0/lib64:/lib
export PATH=/usr/local/cuda-5.0/bin:$PATH
IT purhcased a Drobo backup box for the Ubuntu server (\\10.1.10.163\BSimison). It is a self-contained box running a Drobo
proprietary file format called BeyondRAID. The system supports CIFS, SMB, and AFP.
The following is how I mounted the Drobo drive:
First I made a directory as a mountpoint.
$ mkdir -p /backup/Drobo
then I mounted the networked Drobo box to that mountpoint
$ sudo mount -t cifs //10.1.10.163/BSimison /backup/Drobo -o username=*****,password=*****,rw,uid=*****,gid=*****
The 'username' and 'password' are my credentials assigned to the Drobo box, 'uid' and 'gid' are my user id and group id on the Ubuntu box.
I then added this mount to the etc/fstab so that it would automatically mount the Drobo box on reboots.
First backed up the fstab file:
$ sudo cp /etc/fstab /ect/fstab.bak
I then edited the fstab with the folllowing lines at the end of the file:
# mounting networked Drobo backup drive to /backup/Drobo
10.1.10.163:/BSimison /backup/Drobo cifs -o rw,nolock,root 1 2
To back up a specific MySQL database I have only had success with the phpMyAdmin web interface. The command line option of:$ mysqldump -u xxxx -p xxxx embiotocid_1_radtags > embiotocid_1_radtags.sq
throws the error:$ mysqldump: unknown variable 'local-infile=1'
I tried numerous options recommended by various online help forums, but none worked. Including deleting the local-infile=1 line in /usr/local/share/stacks/sql/mysql.cnf and restarting MySQL with:$ sudo service mysql restart
but got the same error. I was able to log into phpMyAdmin, select the database I wanted to export (backup), select all tables, select the "Export" tab and then selected the "Quick - display only the minimal options" option.
The system froze while "ejecting" a MacOS formatted external drive connected via USB. Upon restarting, the GUI login screen did
not load; only a terminal login option was presented. The error message inlcuded, among many other things, that the NVIDIA
drivers did not match. I tried to restore an old /etc/X11/xorg.conf file and rebooted with no success. I then tried to manually start
an X session with: $ sudo startx
This failed to start the X session.
I then tried to restart the lightdm service (X display manager) with: $ sudo service lightdm start
Eventually, I decided to reinstall the NVIDIA drivers (really should be a last resort option). First updating apt-get:
$ sudo apt-get update
$ sudo apt-get install nvidia-current
This resolved the problem.
To install an ssh daemon so users can use the ssh protocol to access the Ubuntu server I simply used the apt-get command:$ sudo apt-get install openssh-server
This starts the ssh daemon and configures the init.d files such that it starts with boot up.