update retry-on-failed acceptance docs

Relaunching by other status values is tabled for later.

awx/api/serializers.py

@@ -2648,8 +2648,7 @@ class JobRelaunchSerializer(JobSerializer):
         required=False, allow_null=True, default='all',
         choices=[
             ('all', _('No change to job limit')),
-            ('failed', _('All failed and unreachable hosts')),
+            ('failed', _('All failed and unreachable hosts'))
-            ('unreachable', _('Unreachable hosts'))
         ],
         write_only=True
     )

docs/retry_by_status.md

@@ -1,7 +1,16 @@
 # Relaunch on Hosts with Status
 
-This feature allows the user to relaunch a job, targeting only a subset
+This feature allows the user to relaunch a job, targeting only hosts marked
-of hosts that had a particular status in the prior job.
+as failed in the original job.
+
+### Definition of "failed"
+
+This feature will relaunch against "failed hosts" in the original job, which
+is different from "hosts with failed tasks". Unreachable hosts can have
+no failed tasks. This means that the count of "failed hosts" can be different
+from the failed count, given in the summary at the end of a playbook.
+
+This definition corresponds to Ansible .retry files.
 
 ### API Design of Relaunch
 
@@ -11,25 +20,17 @@ POST to `/api/v2/jobs/N/relaunch/` without any request data should relaunch
 the job with the same `limit` value that the original job used, which
 may be an empty string.
 
+This is implicitly the "all" option below.
+
 #### Relaunch by Status
 
-Providing request data containing `{"hosts": "<status>"}` should change
+Providing request data containing `{"hosts": "failed"}` should change
-the `limit` of the relaunched job to target the hosts matching that status
+the `limit` of the relaunched job to target failed hosts from the previous
-from the previous job (unless the default option of "all" is used).
+job. Hosts will be provided as a comma-separated list in the limit. Formally,
-The options and meanings of `<status>` include:
+these are options
 
  - all: relaunch without changing the job limit
- - ok: relaunch against all hosts with >=1 tasks that returned the "ok" status
+ - failed: relaunch against all hos
- - changed: relaunch against all hosts with >=1 tasks had a changed status
- - failed: relaunch against all hosts with >=1 tasks failed plus all unreachable hosts
- - unreachable: relaunch against all hosts with >=1 task when they were unreachable
-
-These correspond to the playbook summary states from a playbook run, with
-the notable exception of "failed" hosts. Ansible does not count an unreachable
-event as a failed task, so unreachable hosts can (and often do) have no
-associated failed tasks. The "failed" status here will still target both
-status types, because Ansible will mark the _host_ as failed and include it
-in the retry file if it was unreachable.
 
 ### Relaunch Endpoint
 
@@ -41,16 +42,26 @@ regarding the host summary of the last job. Example response:
     "passwords_needed_to_start": [],
     "retry_counts": {
         "all": 30,
-        "failed": 18,
+        "failed": 18
-        "ok": 25,
-        "changed": 4,
-        "unreachable": 9
     }
 }
 ```
 
 If the user launches, providing a status for which there were 0 hosts,
-then the request will be rejected.
+then the request will be rejected. For example, if a GET yielded:
+
+```json
+{
+    "passwords_needed_to_start": [],
+    "retry_counts": {
+        "all": 30,
+        "failed": 0
+    }
+}
+```
+
+Then a POST of `{"hosts": "failed"}` should return a descriptive response
+with a 400-level status code.
 
 # Acceptance Criteria