On Sat, Sep 24, 2011 at 4:34 PM, Chris Covington <[email protected]> wrote: > hi all, > just wanted to say i love parallel and what it has done on single machines > for me.
If you like GNU Parallel: * Post the intro videos on forums/blogs/Twitter/Facebook/Linkedin * Request or write a review for your favourite blog or magazine * Invite me as a speaker for your next conference If GNU Parallel saves you money: * (Have your company) donate to FSF https://my.fsf.org/donate/ > my question is about the remote portion which i'm excited about > using as well. > the catch is in the environment that i'll be using parallel: > 1) parallel will be available with the same path on every machine (ie > /foo/bar/parallel) , but /foo/bar/parallel will not be in the $PATH. right > now parallel just sshes to the remote boxes and then it fails because > 'parallel' is not found. You do not need GNU Parallel installed on the remote machines. It is used for figuring out the number of CPUs, which you can pass to it using the '/' notation. This command will not call parallel on my4coremachine: parallel -S 4/my4coremachine.example.net echo ::: bar > 2) the user that parallel runs as will have a read-only homedir. Since you do not need to call GNU Parallel on the remote machines I reckon this will not be an issue. The .parallel dir is used for small files (such as caching loadavg if using --load) but not for storing the output from the commands. The output from the commands is stored in $TMPDIR. See --tmpdir. I have not run a thorough test, but is seems most functionality will work fine with a readonly homedir. If you do make a thorough test it might be interesting to see what functionality is missing with a read-only homedir. /Ole
