In fabric, I have a task collecting something on a per host basis (small example).
from fabric.api import task, run, hide
env.hosts['h1', 'h2', 'h3']
@task
def info():
with hide('everything'):
info = run("who | tail -n 1")
print("On host {0} last user was {1}".format(env.host_string, info))
Run with
fab info
will give something like
[h1] Executing task 'info'
On host h1 last user was userXX pts/29 2015-07-29 15:57 (:0)
[h2] Executing task 'info'
On host h2 last user was userXX pts/29 2015-07-29 16:57 (:0)
[h3] Executing task 'info'
On host h3 last user was userXX pts/29 2015-07-29 17:57 (:0)
While this is fine for 3 or 5 hosts, it gets very hard to view for 20 or more hosts (or more complicated output). What I want to do is to accumulate all output for each host and use this to create a summary/overview in the end, after the task was executed on each host.
How would I do that?
... while writing this question and after at least an hour of googleing different phrases, I finally found this:
http://docs.fabfile.org/en/1.14/usage/execution.html#leveraging-execute-to-access-multi-host-results
Even though I scanned through this site a couple of times I missed the relevant part, and therefore thought to post it here with the hope its easier to find if one does not search for the exact phrase on google:
from fabric.api import task, execute, run, runs_once
@task
def workhorse():
return run("get my infos")
@task
@runs_once
def go():
results = execute(workhorse)
print results
So the example from the question can be solved with:
from fabric.api import task, run, hide, execute, runs_once
env.hosts['h1', 'h2', 'h3']
@task
def collect_info():
with hide('everything'):
info = run("who | tail -n 1")
return info
@task
@runs_once
def info():
collected_output = execute(collect_info)
for host, info in collected_output.iteritems():
print("On host {0} last user was {1}".format(host, info))