gistfile1.py

=== modified file 'schema.py'
--- schema.py  2012-12-05 15:38:38 +0000
+++ schema.py	2013-02-05 21:02:18 +0000
@@ -63,6 +63,9 @@
         if 'BadRequest' not in cfs:
             workaround_1779(mgr.create_column_family, keyspace, 'BadRequest',
                 default_validation_class=CounterColumnType())
+        if 'BucketSystems' not in cfs:
+            workaround_1779(mgr.create_column_family, keyspace, 'BucketSystems',
+                default_validation_class=UTF8_TYPE)
     finally:
         mgr.close()
 

=== modified file 'submit.wsgi'
--- submit.wsgi	2013-01-16 18:07:15 +0000
+++ submit.wsgi	2013-02-04 22:07:25 +0000
@@ -121,6 +121,7 @@
     package, version = utils.split_package_and_version(package)
     src_package, src_version = utils.split_package_and_version(src_package)
     fields = utils.get_fields_for_bucket_counters(problem_type, release, package, version)
+    # user_token is a misnomer as it is really system specific
     if user_token:
         data['SystemIdentifier'] = user_token
     oopses.insert_dict(oops_config, oops_id, data, user_token, fields)

=== modified file 'utils.py'
--- utils.py	2012-11-30 10:38:54 +0000
+++ utils.py	2013-02-06 15:25:43 +0000
@@ -1,10 +1,32 @@
 from oops import Config
 from oops_wsgi import install_hooks, make_app
 from oops_datedir_repo import DateDirRepo, serializer_rfc822
+from pycassa.cassandra.ttypes import NotFoundException
 
 from oopsrepository import oopses
 import apt
 import os
+import pycassa
+
+configuration = None
+try:
+    import local_config as configuration
+except ImportError:
+    pass
+if not configuration:
+    import configuration
+
+import metrics
+class FailureListener(pycassa.pool.PoolListener):
+    def connection_failed(self, dic):
+        name = 'cassandra_connection_failures'
+        metrics.get_metrics().increment(name)
+
+creds = {'username': configuration.cassandra_username,
+         'password': configuration.cassandra_password}
+pool = pycassa.ConnectionPool(configuration.cassandra_keyspace,
+                              configuration.cassandra_hosts,
+                              listeners=[FailureListener()], credentials=creds)
 
 def get_fields_for_bucket_counters(problem_type, release, package, version):
     fields = []
@@ -39,10 +61,13 @@
     return (package, version)
 
 def bucket(oops_config, oops_id, crash_signature, report_dict):
+    bucketversions_cf = pycassa.ColumnFamily(pool, 'BucketVersions')
+    bucketsystems_cf = pycassa.ColumnFamily(pool, 'BucketSystems')
     release = report_dict.get('DistroRelease', '')
     package = report_dict.get('Package', '')
     problem_type = report_dict.get('ProblemType', '')
     dependencies = report_dict.get('Dependencies', '')
+    system_uuid = report_dict.get('SystemIdentifier', '')
     if '[origin:' in package or '[origin:' in dependencies:
         # This package came from a third-party source. We do not want to show
         # its version as the Last Seen field on the most common problems table,
@@ -56,6 +81,22 @@
         package, version = split_package_and_version(package)
 
     fields = get_fields_for_bucket_counters(problem_type, release, package, version)
+    try:
+        bucket_versions = bucketversions_cf.get(crash_signature)
+        first_version, version_count = sorted(bucket_versions.iteritems(),
+            cmp=apt.apt_pkg.version_compare, key=lambda t: t[0])[0]
+    except NotFoundException:
+        # it doesn't exist in bucketversions so we want to create it
+        first_version = version
+        version_count = 0
+    # if version count is high it must have affected a lot of systems right?
+    if version == first_version and version_count < 8:
+        try:
+            systems = bucketsystems_cf.get(crash_signature)
+	    if system_uuid not in systems:
+                bucketsystems_cf.insert(crash_signature, {system_uuid: ''})
+        except NotFoundException:
+            bucketsystems_cf.insert(crash_signature, {system_uuid: ''})
     oopses.bucket(oops_config, oops_id, crash_signature, fields)
     if (package and version) and not third_party:
         oopses.update_bucket_metadata(oops_config, crash_signature, package,