Troubleshooting

Installation issues

Follow the Installation instructions to install naarad and its necessary prerequisites. We have found that the steps below should help address most issues encountered during installation of dependencies, specially matplotlib.

Python version

If there are multiple python versions on your system, you should pick a version to work with and use pip for that version.

Let's say you pick python 2.6, then make sure to run pip2.6 when installing requirements.

Pip version

Some machines may have multiple versions of pip installed. If you are facing problems with some naarad requirements (e.g., if pip says requirement is satisfied, but naarad still doesn't work), make sure that you are using the "system" version of pip, usually in /usr/local/bin/pip.

$ which pip
/usr/local/bin/pip

distribute / setuptools issue

Ensure setuptools/distribute are up to date. As sudo/root,

easy_install -U distribute
easy_install -U setuptools
pip install --upgrade distribute
pip install --upgrade setuptools

Numpy installation issues

If pip install numpy fails, most likely you need python development packages from python-dev or python-devel.

On Redhat,

sudo yum install python-devel

On Ubuntu,

sudo apt-get install python-dev

Matplotlib installation issues

In general, most issues with naarad installation are due to matplotlib. Here are some steps below to fix matplotlib installation and other general installation issues. If there are still issues installing matplotlib, check out their installation page.

Linux

On Redhat,

sudo yum install gcc gcc-c++ python-devel freetype-devel libpng-devel libxml2-devel libxslt-devel pip

On Ubuntu,

sudo apt-get install gcc python-dev libfreetype6-dev libpng12-dev

OSX

Install XQuartz (http://xquartz.macosforge.org/landing/) to satisfy freetype dependency for matplotlib.

Package versions

List versions of installed python dependencies for naarad. Versions should match or exceed whats listed in requirements.txt

pip list | grep -e numpy -e matplotlib -e pygal -e lxml -e argparse -e lxml -e pyparsing -e pytz
argparse (1.1)
pyparsing (2.0.1)
pytz (2013.8)
lxml (3.2.4)
pygal (1.2.0)
numpy (1.7.0)
matplotlib (1.3.1)

If the package versions are lower than listed above, use the following commands to upgrade them:

pip install --upgrade pyparsing
pip install --upgrade numpy
pip install --upgrade matplotlib
pip install --upgrade lxml
pip install --upgrade pygal
pip install --upgrade argparse
pip install --upgrade pytz

Runtime Questions

Q. naarad does not recognize the timestamp in my data files.

A. You can find the list of timestamp formats naarad currently supports @ https://github.com/linkedin/naarad/wiki/Supported-Timestamp-Formats

Javascript Charting Issues

Q. Charting does not work when I load naarad reports locally from the filesystem in Google Chrome?

A. Google Chrome by default blocks ajax calls to file resources. The workaround is to start Chrome with --allow-file-access-from-files parameter.

A. You can also go into the naarad_output folder and running:

python -m SimpleHTTPServer

This will host a simple HTTP server in the directory so that you can then browse to localhost:8000/report.html or localhost:8000/summary.html for the reports and the loading will work.