Noisier failures with ssh-group

[tomconnors]

The docs indicate that if one server in ssh-group has a failure, the others will continue. I think that should be configurable (I'd like the whole thing to crash in that case), but it should at least log a warning. The following code just logs "I'm on the server" then :ok

(ns test
  (:require
   [spire.modules :as spire]
   [clojure.string :as string]))

(defn main []
  (spire/ssh-group
   [{:hostname "10.10.3.250" :username "ubuntu"}]
   (spire/debug "I'm on the server")
   (spire/debug (string/trim nil))
   (spire/debug "All done"))
  :ok)

(main)

[retrogradeorbit]

This could be done. The confusing issue is that one host may be much further along the script than the failing host. Or much earlier. The hosts don't run in lockstep. They each run as fast as they can. Is that an issue for what you are doing? Is it just about error readability?

[tomconnors]

This is just about readability; I had a failing script and didn't realize it.

[retrogradeorbit]

What do you think of collating those errors and reporting them at the end of execution?

[tomconnors]

That would work for me.