Installation of KAGRA summary pages

Writer : Duncan

ROOT set up

The following should be performed as the root user

Add Debian repositories as here
Install ssh-server:
```
   1  apt-get update
   2  apt-get install openssh-server
   3  systemctl start ssh
```
Install Apache (for web server):
```
   1  apt-get install apache2
   2  a2enmod userdir
   3  systemctl restart apache2
```
To enable .htaccess files within the userdir configuration for Apache, the all option was added to the <Directory /home/*/public_html> AllowOverride directive as follows:
```
   1  <IfModule mod_userdir.c>
   2  ...
   3      <Directory /home/*/public_html>
   4          AllowOverride FileInfo AuthConfig Limit Indexes AllowOverride
   5              ...
   6      </Directory>
   7  </IfModule>
```
Add LIGO Debian repositories as here, and install keyring to trust:
```
   1  apt-get install lscsoft-archive-keyring
```

Installed basic dependencies:

   1  apt-get install \
   2      git \
   3      python-pip \
   4      python-virtualenv \
   5      python-nds2-client \
   6      lal-python \
   7      ldas-tools-framecpp-python \
   8      python-gwpy \
   9      lalapps

Summary pages setup

The following is performed as the controls user
Set up a virtualenv for the summary page code:
```
   1  python -m virtualenv ~/opt/summary-2.7 --system-site-packages
```
Activate the virtualenv to enter that environment:
```
   1  source ~/opt/summary-2.7/bin/activate
```
And then install everything we need:
```
   1  python -m pip install \
   2      "gwpy>=0.12.0"
   3      git+https://github.com/gwpy/gwsumm.git
```

Automation setup

Install HTCondor:
```
   1  apt-get install htcondor
```

Automation notes

The automation is handled using HTCondor's timescheduling system.

If the Condor queue is empty, you need to restart the scheduled jobs:
```
   1  cd /home/controls/etc/summary/condor
   2  condor_submit gw_daily_summary_kagra.sub
   3  condor_submit gw_daily_summary_rerun_kagra.sub
```
these jobs will persist in the Condor queue. If the persistens jobs are held, you cannot condor_release them, you must condor_rm and condor_submit again. For any other jobs (gw_summary), condor_release works fine

Monitoring setup

I (Duncan) have installed Ganglia on k1sum0 to allow web-based monitoring of the CPU/memory/disk/network etc. This is achieved as follows:

Install ganglia software:
```
   1  apt-get install ganglia-monitor gmetad ganglia-webfrontend
```
Link Ganglia Apache configuration and restart:
```
   1  ln -s /etc/ganglia-webfrontend/apache.conf /etc/apache2/sites-enabled/ganglia.conf
```

Edit configuration files to include some useful information:

 /etc/ganglia/gmond.conf  # edit cluster parameters around line 20
 /etc/ganglia/gmetad.conf  # edit gridname on line 72

Restart all of the necessary services
```
   1  systemctl restart ganglia-monitor gmetad apache2
```

The Ganglia output is viewable on http://k1sum0/ganglia/.

Ganglia is designed to monitor an entire cluster of machines; this page might be useful if you wish to add other machines to the Ganglia output.

After Installation

Writer : TYamamoto

After the installation of gwpy and the daily-summary tools, TYo installed emacs.

> su 
> apt-get install emacs

After the installation of gwpy and the daily-summary tools, TY installed sudo and nfs-common. nfs-common is required we use nfs as the mount type on /etc/fstab.

> su -
> apt-get install sudo nfs-common
> emacs -nw /etc/fstab
k1nfs0:/export/users       /users       nfs            rw,bg,soft  0 0
> mount -a

KAGRA/Subgroups/PEM/SummaryPage/k1sum0 (last edited 2018-10-19 16:01:29 by duncan.macleod)