If you are going to run the tests on the low-memory host overs ssh, why do you need the slave? Couldn't the master execute the ssh?
Also, I myself have executed tests on a Jenkins slave with 650 MB of RAM (a virtual machine), so your statement about the 2 GB requirement seems high. But I admit, it depends on what kind of plugins you have loaded into Jenkins and what kind of test result XML files you want Jenkins to read and parse. -- Sami Gooze <jmpo...@gmail.com> kirjoitti 27.4.2012 kello 15.09: > Dear all, > > OpenSC free software project is using Jenkins to make software > releases and compile on a farm and would like to thank the community > for their hard word. > > Just one question: we would like to automate a number of regression > tests on slaves. To be more precise, we would like to install OpenSC > latest packages and run a number of regression tests. These regression > tests are simple commands. > > The problem is that our test show that a Jenkins slave should have at > least 2Gb memory to work. But we would like to use PCEngines embedded > boards for testing. These board have only 256Mb onboard memory. This > is enough to run GNU/Linux and FreeBSD, but not start a real Jenkins > slave. > > So what would you suggest to be able to connect to low memory clients > without starting a slave? Shall we simple run a command on a slave > with 2Gb memory with SSH command to execute remotely anywhere? Or is > there a more elegant solution available as a plugin? > > Kind regards, > Jean-Michel Pouré > GOOZE smartcards > http://www.gooze.eu