I was wondering if there are best practices for backing up a Hudson instance. The Hudson book only suggests backing up $HUDSON_HOME and lists some plugins that I'd rather not use.
For our use case (more about it ), I don't want to back up all of $HUDSON_HOME because there are rather large workspaces that take lots of unnecessary space in the backup. Also, restore has to be quick because we don't have an HA setup. Most backup tools support excluding certain directories, but since workspaces are inside the job directories this makes things more complicated than necessary.
Is there a way to move the workspaces out of $HUDSON_HOME (an environment variable $HUDSON_WORKSPACES would be ideal)? Right now, our solution is to not build on the master at all and instead run a build slave on the same machine. It works, but it feels like a hack.
Thanks in advance,
 http://blog.mafr.de/2011/12/27/hudson-vs-jenkins-revisited/ (sorry, apparently I'm not allowed to post links)
Personally I use the ThinBackup plugin which scrapes the correct bits out of Hudson_Home, however, that been said I see where you're coming from. It makes eminent sense to provide the option to separate configuration from the working data. As far as I'm aware there is no way to do this right now (although I could be wrong)
Anyway, I've logged an enhancement on this one bugs.eclipse.org/bugs/show_bug.cgi?id=374342
It's a good thing to discuss in the next community meeting as well I think that it would garner a lot of support. I've added it to the agenda there.
Thanks for your reply and also for creating the enhancement request! BTW, this feature would also make it easier to sync HUDSON_HOME to a hot standby machine using DRBD or something similar, taking us a step further towards an HA setup.