erikfrey / bashreduce Goto Github PK

In README, it is said that we can use bashreduce as a poorman's dfs by doing

br -r "cat > /tmp/myfile" < input

I think I can see a file 'input' will be distributed among the hosts with the file name /tmp/myfile.
But how can we read the file?

br -r "cat /tmp/myfile"

would give us a broken file since those distributed chunks do not have information about in what order the original file was written.
Or it is assumed that input is a file whose lines have numbers in order?

Any luck with the redhat system? I tried removing the -q0 from nc and that took me a little further.

However, there seem to be other problems with the ssh'ing.

i have run this command to test it:

br < Makefile

and get nothing...

I don't get how I should define the hosts in my configuration file.

I have access to the following hosts:

• gioby@localhost
• gdallolio@server1
• gioby@server2

I have tried to run bashreduce with the following command line:
$: br -m "gioby@localhost gdallolio@server1" -r 'head' < bigfile.txt
but it asks me the password to connect to localhost.

Could you improve the documentation on how to specifiy hosts, with different usernames?

(I'm working on this one — if I find something I'll send a pull request. Consider this as much a development log as a feature request!)

Apple's version of nc doesn't support -q, though -w seems to do something similar. Even with the change, it results in the rather mystifying behavior of reading the input and producing no output, and I have no idea why.

(This is of course after installing pv from Homebrew; one could easily alias pv=cat if one was so inclined!)

on first run pv was not found

sudo apt-get install pv

Hi there,

I just started reading the README and was wondering about the first table (comparsing sort to br). What is the difference between the second and third row? It looks the same to me.