Bug 1254673 - Stop posting new resultset info to pulse

schemas/resultset-message.json

@@ -1,30 +0,0 @@
-{
-  "id":           "https://treeherder.mozilla.org/schemas/v1/resultset-message.json#",
-  "$schema":      "http://json-schema.org/draft-04/schema#",
-  "title":        "New ResultSet Message",
-  "description":  "Pulse message sent whenever a new result-set is created.",
-  "type":         "object",
-  "properties": {
-    "version": {
-      "title":                "Message-format version",
-      "enum":                 [1]
-    },
-    "project": {
-      "title":                "Project Name",
-      "description":          "Identifier for treeherder project, like `try` or `mozilla-central`.",
-      "type":                 "string"
-    },
-    "revision_hash": {
-      "title":                "Revision Hash Identifier",
-      "description":          "Identifier for the result-set that was created.",
-      "type":                 "string"
-    },
-    "repository_url": {
-      "title":                "Repository URL",
-      "description":          "URL for the repository for the revision.",
-      "type":                 "string"
-    }
-  },
-  "additionalProperties":     true,
-  "required": ["version", "revision_hash"]
-}

tests/conftest.py

@@ -353,11 +353,6 @@ def pulse_consumer(exchange, request):
     return simpleQueue
 
 
-@pytest.fixture
-def pulse_resultset_consumer(request):
-    return pulse_consumer('new-result-set', request)
-
-
 @pytest.fixture
 def pulse_action_consumer(request):
     return pulse_consumer('job-actions', request)

tests/etl/test_pushlog.py

@@ -11,7 +11,7 @@ from treeherder.etl.pushlog import (HgPushlogProcess,
 
 def test_ingest_hg_pushlog(jm, initial_data, test_base_dir,
                            test_repository, mock_post_json,
-                           activate_responses, pulse_resultset_consumer):
+                           activate_responses):
     """ingesting a number of pushes should populate result set and revisions"""
 
     pushlog_path = os.path.join(test_base_dir, 'sample_data', 'hg_pushlog.json')
@@ -34,19 +34,6 @@ def test_ingest_hg_pushlog(jm, initial_data, test_base_dir,
 
     assert len(pushes_stored) == push_num
 
-    rev_to_push = set()
-    for push in json.loads(pushlog_content)['pushes'].values():
-        # Add each rev to the set remember we shorten them all down to 12 chars
-        rev_to_push.add(push['changesets'][-1]['node'][0:12])
-
-    # Ensure for each push we sent a pulse notification...
-    for _ in range(0, push_num):
-        message = pulse_resultset_consumer.get(block=True, timeout=2)
-        content = message.payload
-        assert content['revision'] in rev_to_push
-        # Ensure we don't match the same revision twice...
-        rev_to_push.remove(content['revision'])
-
     revisions_stored = jm.get_dhub().execute(
         proc="jobs_test.selects.revision_ids",
         return_type='tuple'

tests/model/derived/test_refdata.py

@@ -258,27 +258,6 @@ def test_add_job_type(refdata):
     assert row_data == expected
 
 
-def test_get_repository_info(refdata, repository_id):
-    """test get_repository_info retrieves the right informations"""
-
-    info = refdata.get_repository_info(repository_id)
-
-    expected = {
-        "dvcs_type": "hg",
-        "name": "mozilla-central",
-        "url": "https://hg.mozilla.org/mozilla-central",
-        "active_status": "active",
-        "codebase": "gecko",
-        "repository_group_id": 1,
-        "description": ""
-    }
-
-    refdata.disconnect()
-
-    for k, v in expected.items():
-        assert info[k] == v
-
-
 @pytest.fixture
 def sample_bugs(test_base_dir):
     filename = os.path.join(

treeherder/model/derived/jobs.py

@@ -17,7 +17,6 @@ from treeherder.model.models import (Datasource,
                                      Repository)
 from treeherder.model.tasks import (populate_error_summary,
                                     publish_job_action,
-                                    publish_resultset,
                                     publish_resultset_action)
 
 from .artifacts import ArtifactsModel

treeherder/model/derived/refdata.py

@@ -1161,18 +1161,6 @@ class RefDataManager(object):
 
         return id_iter.get_column_data('id')
 
-    def get_repository_info(self, repository_id):
-        """retrieves all the attributes of a repository"""
-
-        repo = self.execute(
-            proc='reference.selects.get_repository_info',
-            placeholders=[repository_id],
-            debug_show=self.DEBUG,
-            return_type='iter')
-        # retrieve the first elem from DataIterator
-        for r in repo:
-            return r
-
     def get_all_repository_info(self):
         return self.execute(
             proc='reference.selects.get_all_repository_info',

treeherder/model/exchanges.py

@@ -10,26 +10,6 @@ class TreeherderPublisher(PulsePublisher):
     """
     exchange_prefix = "v1/"
 
-    new_result_set = Exchange(
-        exchange='new-result-set',
-        title="New Result-Set Messages",
-        description="""
-            Whenever a new result-set is created a message featuring the
-            `revision_hash` is published on this exchange.
-        """,
-        routing_keys=[
-            Key(
-                name='project',
-                summary="Project (or branch) that this result-set concerns"
-            ),
-            Key(
-                name='revision_hash',
-                summary="result-set identifier for the message"
-            )
-        ],
-        schema="https://treeherder.mozilla.org/schemas/v1/resultset-message.json#"
-    )
-
     resultset_action = Exchange(
         exchange="resultset-actions",
         title="Actions issued by resultset",

treeherder/model/sql/reference.json

@@ -203,19 +203,6 @@
                     WHERE `name` = ?",
             "host_type":"read_host"
         },
-        "get_repository_info":{
-            "sql": "SELECT *
-                    FROM `repository`
-                    WHERE `id` = ?  AND `active_status` = 'active'",
-            "host_type":"read_host"
-        },
-        "get_all_repository_info":{
-            "sql": "SELECT *
-                    FROM `repository`
-                    WHERE
-                        `active_status` = 'active'",
-            "host_type":"read_host"
-        },
         "get_all_option_collections":{
             "sql":"SELECT option_collection_hash,
                         GROUP_CONCAT( name SEPARATOR ' ' ) as opt

treeherder/model/tasks.py

@@ -124,52 +124,6 @@ def publish_resultset_runnable_job_action(project, resultset_id, requester,
     )
 
 
-@task(name='publish-resultset')
-def publish_resultset(project, ids):
-    # If we don't have a publisher (because of missing configs), then we can't
-    # publish any pulse messages. This is okay, local installs etc. doesn't
-    # need to publish on pulse, and requiring a pulse user is adding more
-    # overhead to an already large development setup process.
-    publisher = pulse_connection.get_publisher()
-    if not publisher:
-        return
-
-    from treeherder.model.derived.jobs import JobsModel
-
-    with JobsModel(project) as jm:
-        # Publish messages with new result-sets
-        for entry in jm.get_result_set_list_by_ids(ids):
-            repository = jm.refdata_model.get_repository_info(entry['repository_id'])
-
-            if repository is None:
-                return
-
-            entry['repository_url'] = repository['url']
-
-            # Don't expose these properties, they are internal, at least that's
-            # what I think without documentation I have no clue... what any of
-            # this is
-            del entry['revisions']      # Not really internal, but too big
-            del entry['repository_id']
-
-            # Set required properties
-            entry['version'] = 1
-            entry['project'] = project
-            # Property revision_hash should already be there, I suspect it is the
-            # result-set identifier...
-
-            # publish the data to pulse
-            publisher.new_result_set(**entry)
-
-            # Basically, I have no idea what context this runs and was inherently
-            # unable to make kombu with or without pyamqp, etc. confirm-publish,
-            # so we're stuck with this super ugly hack where we just close the
-            # connection so that if the process context is destroyed then at least
-            # messages will still get published... Well, assuming nothing goes
-            # wrong, because we're not using confirm channels for publishing...
-            publisher.connection.release()
-
-
 @task(name='populate-error-summary')
 def populate_error_summary(project, artifacts, job_id_lookup):
     """