Skip to content. | Skip to navigation

Personal tools


You are here: Home / Wiki / Perexperimentdb


Per Experiment Databases

Per Experiment Databases

Per Experiment Databases are intended to make it easy for experimenters to store and manage data, without having to worry about installing mysql on their images, or backing up their databases (since the disks on experimental nodes are not filesaved). Instead, each experiment gets it own mysql database on users (ops) when the experiment is first created. All of the users in the project (or group) have access to that database.

Per Experiment Databases are controlled by a directive in your NS file when you create your experiment. To make sure that the database is created and the proper access granted, add this to your NS file.

	tb-set-dpdb 1

A quick rundown of important details.

  • Each experiment gets its own DB on ops, named by "$pid+$eid"
  • Each user gets a DB account on ops, which you can access from ops or your experimental nodes with mysql, or any of the other mysql APIs (perl, python, etc).
  • Your mysql password is the same as your mailmain,jabber,wikipassword, which you can find by looking on the My Chat Buddies page in the Collaboration drop down menu in the web interface. There is also an XMLRPC interface to get your password (the user.collabpassword) method.
  • To access an experiment DB on ops, you can do this:
    	mysql -u stoller -pPASSWORD testbed+myexp
  • To access an experiment DB from an experimental node, do this:
    	mysql -h ops -u stoller -pPASSWORD testbed+myexp