screen for cluster management

After my parallel ssh roundup and hacking my own solution...

I just realized screen can do this too using "at" and "stuff".  Here's how:

1. open a bunch of screens and ssh to various hosts
2. get to a screen command prompt (control key followed by colon, ^C-a : )
3. at the prompt, enter the command "at \#", this tells screen to run command that follows on all windows (yes, include the backslash and don't press enter yet)
4. continue with the command "stuff", which tells screen to quite literally stuff whatever follows it into the input of a screen terminal (don't press enter quite yet)
5. follow stuff with the command you want to run, you'll need to put it in quotes if the command has arguments (not yet)
6. lastly, you need to end the line with a return so the shell(s) will run the command, so end the line with "\012" which is octal for a newline character

Now, press enter and what your command get run across all of your screen sessions.  This is particular cool for long running commands where you can switch back and forth between screen windows and check on the progress of many machines or also useful if you have a couple of quick maintenance task to run on a bunch of hosts.

So, putting it all together, the whole sequence looks like this:

^C-a : at \# stuff ls \012
^C-a : at \# stuff 'ls -alR' \012

broadcast ssh - my entry into the world of parallel or cluster ssh tools

I threatened before when writing up my review of the parallel or cluster ssh tools out there that I would write my own. And after a quick review of paramiko, especially the demo scripts included, this turned out to be a pretty quick hack.

The basics of my ssh client are this:

• prompt for username and password - ssh keys not required (although I would like to add support for using them if available in the future)
• interactive use - I often want to look at things, which leads me to want to look at other things; in other words, I don't want to be constrained by having a list of commands to execute up front
• parallel - I have a bunch of commands and a bunch of machines, a simple loop executing each command on each machine isn't going to cut it

Those last two together make things a little tricky, the interactive bit means you have two choices.  You can be line oriented, prompt for a command and send it to each host or you can be completely interactive and send each character.

I started with the former, but soon found that if you only send commands, you have practically no environment setup (that is all done in the shell, remember...).  So, despite the difficulties in dealing with the terminal, that is what this implementation does.

A terminal based solution has its own issues, the line buffering of output from each host has to be dealt with to avoid interleaved garbage as the results of each command.  But in the end that wasn't too hard.

In the end, usage is simple, cut-n-paste the code, save it as bssh.py and run something like:

./bssh.py host1 host2 host3 host4

Enter a username and password at the prompts, and away you go, just enter commands as you would with any other shell. You will of course need paramiko and its dependency pycrypto installed and available.

parallel ssh tool roundup

So I'm in the market for a good "parallel" ssh tool.  Basically, I want to ssh and type some commands like I always do except that instead of one command output, I want the command to be run on a bunch of machines and I want responses from each.  I've used things like mpiexec in the past, but I was hoping for something more ad-hoc.  I just want to specify hosts on the command line (with no prior setup).  I really don't even want to required ssh keys or having the same password if I can avoid it.  Like I said, just ssh as usual, run some commands, get multiple results.

Anyway, these are the tools I've come across (in no particular order).  I'll outline what I regard as the advantages and disadvantages of each below.

1. pdsh - "a high-performance, parallel remote shell utility" from Lawrence Livermore National Laboratory
2. pssh an implementation of some parallel ssh tools in python
3. dsh - dancer's / distributed shell
4. pydsh - a python version of dancer's shell
5. clusterssh - "a tool for making the same change on multiple servers"
6. mussh - "a shell script ... to execute a command or script over ssh on multiple hosts"
7. sshpt - "SSH Power Tool (sshpt) enables you to execute commands and upload files to many servers simultaneously via SSH"
8. multixterm - part of the expect project
9. clusterit - "a collection of clustering tools, to turn your ordinary everyday pile of UNIX workstations into a speedy parallel beast"
10. dish - "The diligence shell 'dish' executes commands via ssh/rsh/telnet/mysql simultaneously on several systems"