For files, its puppet or files only at this stage. (Though packages can handle http so long as the package manager can understand a HTTP file source...)
I started up some discussion on HTTP as a source for files as a means to overcome this. (Linky: http://groups.google.com.au/group/puppet-users/browse_thread/thread/b0d3004ad9daf40c/cf95773e76622eb5 ) Its apparently something that is in the works, and may well go some way towards this under 0.25 when its available... I haven't really looked at the 0.25 betas yet, so I'm not sure how its changed in this regard... For now, I'm making do with an NFS mount under autofs (shudder - Wish I didn't have to do this), and will migrate all the large file serving to run under http once the new release is available and tested... Not ideal, but it means I can do: file { "/nfs/mount/foo.pkg": ensure => present } exec { "/nfs/mount/foo.sh": refreshonly => true, subscribe => File["/nfs/mount/foo.pkg"] } Of course this means that foo.sh gets called each time foo.pkg changes (ie. upgrades...) Greg On Jul 20, 10:32 am, "Fernando Padilla" <f...@alum.mit.edu> wrote: > Thank you. I suppose that's an easy way around it.. I wonder if I want > puppetmaster to also host a simple apache.. > > Or.. does the "source" attribute support http/ftp over just file/puppet > protocols? > > On Sat, 18 Jul 2009 13:13 +0200, "Sylvain Avril" <avr...@gmail.com> > wrote: > > > > > I myself don't use puppet to pull big files. > > Maybe you use puppet with the default Webrick HTTP frontend. You may > > test another frontend like mongrel or passenger : > >http://reductivelabs.com/trac/puppet/wiki/UsingMongrel > >http://reductivelabs.com/trac/puppet/wiki/UsingPassenger > > > For my use, I use an HTTP server and a custom curl definition. But for > > slow connections, it didn't resolve the timeout problem. > > > define common::archive::tar-gz($source, $target) { > > exec {"$name unpack": > > command => "curl ${source} | tar -xzf - -C ${target} && touch > > ${name}", > > creates => $name > > } > > } > > > But the more elegant solution would be to package hadoop. > > > 2009/7/18 Fernando Padilla <f...@alum.mit.edu> > > > > Hi. I'm a beginner, but I have a basic puppet setup working. I am > > > doing a manual tarball installation and it seems to be hanging then > > > eventually timing out on just downloading the file: > > > > file { "/opt/hadoop-0.20.0.tar.gz": > > > source => "puppet:///hadoop020/hadoop-0.20.0.tar.gz" > > > } > > > > I have another module that does the same things and works, my only guess > > > is the size of the tarball: > > > > modules/hadoop020/files/hadoop-0.20.0.tar.gz - 41M > > > modules/zookeeper320/files/zookeeper-3.2.0.tar.gz - 12M > > > > Any ideas or suggestions to speed up file transfers?? > > > > If I manually scp the file, it takes only 30seconds (between office and > > > ec2), why would it take so long and eventually timeout inside the colo ( > > > ec2)? --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Puppet Users" group. To post to this group, send email to puppet-users@googlegroups.com To unsubscribe from this group, send email to puppet-users+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en -~----------~----~----~----~------~----~------~--~---
