From d361f3b44ea06789d4611c413f73a2b4efa5aa77 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ren=C3=A9=20Moser?= <mail@renemoser.net>
Date: Mon, 7 Mar 2016 07:35:04 +0100
Subject: [PATCH] proposal: re-run handlers
---
docs/proposals/re-run-handlers.md | 77 +++++++++++++++++++++++++++++++
1 file changed, 77 insertions(+)
create mode 100644 docs/proposals/re-run-handlers.md
diff --git a/docs/proposals/re-run-handlers.md b/docs/proposals/re-run-handlers.md
new file mode 100644
index 00000000000..9b5a01df8fa
--- /dev/null
+++ b/docs/proposals/re-run-handlers.md
@@ -0,0 +1,77 @@
+# Proposal: Re-run handlers cli option
+
+*Author*: René Moser <@resmo>
+
+*Date*: 07/03/2016
+
+- Status: New
+
+## Motivation
+
+The most annoying thing users face using ansible in production is running handlers manually after a task failed after a notified handler.
+
+### Problems
+
+Handler notifications get lost after a task failed and there is no help from ansible to catch up the notified handlers in a next ansible playbook run.
+
+~~~yaml
+- hosts: localhost
+ gather_facts: no
+ tasks:
+ - name: simple task
+ shell: echo foo
+ notify: get msg out
+
+ - name: this tasks fails
+ fail: msg="something went wrong"
+
+ handlers:
+ - name: get msg out
+ shell: echo handler run
+~~~
+
+Result:
+
+~~~
+$ ansible-playbook test.yml
+
+PLAY ***************************************************************************
+
+TASK [simple task] *************************************************************
+changed: [localhost]
+
+TASK [this tasks fails] ********************************************************
+fatal: [localhost]: FAILED! => {"changed": false, "failed": true, "msg": "something went wrong"}
+
+NO MORE HOSTS LEFT *************************************************************
+
+RUNNING HANDLER [get msg out] **************************************************
+ to retry, use: --limit @test.retry
+
+PLAY RECAP *********************************************************************
+localhost : ok=1 changed=1 unreachable=0 failed=1
+~~~
+
+## Solution proposal
+
+Similar to retry, ansible should provide a way to manully invoke a list of handlers additionaly to the notified handlers in the plays:
+
+~~~
+ $ ansible-playbook test.yml --notify-handlers <handler>,<handler>,<handler>
+ $ ansible-playbook test.yml --notify-handlers @test.handlers
+~~~
+
+Example:
+
+~~~
+ $ ansible-playbook test.yml --notify-handlers "get msg out"
+~~~
+
+The stdout of a failed play should provide an example how to run notified handlers in the next run:
+
+~~~
+...
+RUNNING HANDLER [get msg out] **************************************************
+ to retry, use: --limit @test.retry --notify-handlers @test.handlers
+~~~
+