Ballast: A tool for balancing user load across SSH servers(code.nasa.gov)
code.nasa.gov
Ballast: A tool for balancing user load across SSH servers
http://code.nasa.gov/project/balancing-load-across-systems-ballast/
6 comments
What an odd idea.
Then again, they send stuff into space. Who am I to judge.
Then again, they send stuff into space. Who am I to judge.
When you have enough users, and NFS homedirs, it makes sense.
The University of Waterloo just uses DNS round-robin, but that's not particularly reliable since the load balancing is up to the whims of each client's DNS client implementation.
And then there's software like mosh[0], which, IIRC, does round-robin on the first connection but then sticks to the same server afterwards so that you can e.g. connect to the same pty even if your local IP changed.
[0] http://mosh.mit.edu/
The University of Waterloo just uses DNS round-robin, but that's not particularly reliable since the load balancing is up to the whims of each client's DNS client implementation.
And then there's software like mosh[0], which, IIRC, does round-robin on the first connection but then sticks to the same server afterwards so that you can e.g. connect to the same pty even if your local IP changed.
[0] http://mosh.mit.edu/
> The University of Waterloo just uses DNS round-robin
I take it that they do not allow running persistent applications? I personally would not like to find my screen/tmux session "disappear" after reconnect because I ended to another host.
I take it that they do not allow running persistent applications? I personally would not like to find my screen/tmux session "disappear" after reconnect because I ended to another host.
My University did something that seems similar. There was a group hostname which I believe dropped you into one of a group of linux machines(with nfs homedir) but you cold always specify the actual hostname of the computer where you parked tmux.
Sourceforge? Seriously? Why do NASA people and academia still upload things to SF instead of GitHub?
I'm not sure why they still use Sourceforge, however they did move some of their projects to Github: https://github.com/nasa
This one even has its own separated account: https://github.com/visionworkbench
This one even has its own separated account: https://github.com/visionworkbench
From what I've heard from a friend, probably a few thousand forms and layers of approval to change it.
http://sourceforge.net/projects/ballast/files/ - no sign of a VCS, I wonder what they use internally?
The NASA page claims they use SVN
At a guess, it isn't public - http://svn.code.sf.net/p/ballast/code
Looks like the source is in the distribution files. Download one of these:
http://sourceforge.net/projects/ballast/files/
and untar it and you'll find Perl and C source code.
http://sourceforge.net/projects/ballast/files/
and untar it and you'll find Perl and C source code.
I guess that shows how spoilt I am by GitHub et-al.
Was actually looking at this a few days ago for a cluster of bastion nodes to secured infrastructure. Ts a pretty interesting tool.
Horrible comma use in the description. You don't need either of 'em!
So instead of:
> Ballast is invoked as part of the SSH login process, hence has access to the user name, which is not available in traditional load balancers
You'd write:
> Ballast is invoked as part of the SSH login process hence has access to the user name which is not available in traditional load balancers
I'm afraid i can't agree. Both of those commas seem grammatically necessary to me.
> Ballast is invoked as part of the SSH login process, hence has access to the user name, which is not available in traditional load balancers
You'd write:
> Ballast is invoked as part of the SSH login process hence has access to the user name which is not available in traditional load balancers
I'm afraid i can't agree. Both of those commas seem grammatically necessary to me.
[1] http://www.nas.nasa.gov/hecc/resources/pleiades.html