a lightning quick tutorial on pdsh (for when you need to run the same command on many machines)

All serious Linux administrators have a tool, or few, they use to run commands in mass as it simply isn't practice to ssh into a bunch of machines.  Even for those of us who regularly work only a small number of machines, it is no fun either.  There are many tools out there from free utilities to expensive COTS solutions as well as plenty of RYO specials.  My favorite is pdsh.

The intention of this blog posting is to jump-start anyone who could benefit from learning how to use it.  First, you need to get pdsh installed on your machine; not on all of them, just the server you want to invoke the commands from.  For CentOS/RHEL, you can just run yum install pdsh to get the bits.  Once you have that, you'll want a text file with a list of servers you want to run the commands again such as this example.

n2.hdp22.ha
n3.hdp22.ha
n4.hdp22.ha
n5.hdp22.ha

Then the magic is as simple as the following.

pdsh -R ssh -w ^cluster.txt "<command>"

For an example, what if I wanted to check the current time on a bunch of machines.  I could accomplish that easily enough.

[root@n1 ~]# pdsh -R ssh -w ^cluster.txt "date"
n3: Fri Jan  9 16:51:48 EST 2015
n4: Fri Jan  9 16:51:50 EST 2015
n2: Fri Jan  9 16:51:48 EST 2015
n5: Fri Jan  9 16:51:48 EST 2015
[root@n1 ~]# 

If this was your intro into command blaster utilities, I hope it helps.  If you do this stuff all day, please add a comment with the tool(s) you use.