Not considering replicas for deletions if on certain site

lib/common/interface/mysqlstore.py

@@ -523,6 +523,14 @@ def _do_load_files(self, dataset): #override
 
             dataset.files.add(lfile)
 
+    def _do_check_if_on(self, datasetname, sitename): #override
+        query = 'SELECT COUNT(*) FROM `datasets` AS d'
+        query += ' INNER JOIN `dataset_replicas` AS dr ON dr.`dataset_id` = d.`id`'
+        query += ' INNER JOIN `sites` AS s ON s.`id` = dr.`site_id`'
+        query += ' WHERE  d.`name` = %s and s.`name` = %s'
+
+        return (self._mysql.query(query, datasetname, sitename)[0] != 0)
+
     def _do_find_block_of(self, fullpath, datasets): #override
         query = 'SELECT d.`name`, b.`name` FROM `files` AS f'
         query += ' INNER JOIN `datasets` AS d ON d.`id` = f.`dataset_id`'

lib/common/interface/store.py

@@ -268,6 +268,15 @@ def load_files(self, dataset):
         finally:
             self.release_lock()
 
+    def check_if_on(self, datasetname, sitename):
+        """
+        Return true/false if replica is on specific site.
+        """
+
+        logger.debug('_do_check_if_on()')
+
+        return self._do_check_if_on(datasetname, sitename)
+
     def find_block_of(self, fullpath, datasets):
         """
         Return the Block object for the given file.

lib/detox/configuration.py

@@ -4,5 +4,6 @@
     activity_indicator = '/home/cmsprod/public_html/IntelROCCS/Detox/inActionLock.txt',
     deletion_per_iteration = 0.01, # fraction of quota to delete per iteration
     deletion_volume_per_request = 50, # size to delete per deletion request in TB
+    exclude_if_on = ['T1_IT_CNAF_MSS'], # if a dataset has a replica on these [sites], don't consider it for deletions. Introduced because of CNAF indicent.
     time_shift = 0. # number of days in the future from which to evaluate the policies
 )

lib/detox/main.py

@@ -99,6 +99,16 @@ def _execute_policy(self, policy, is_test, comment):
         # will also select out replicas on sites with quotas
         all_replicas = policy.partition_replicas(self.inventory_manager, target_sites)
 
+        # check if replica is present on other site(s) that trigger exclusion from the possible deletion
+        # possible use case: a tape site has had a water indicident, for example
+        # Communication with database is needed because we do not have all (tape) replicas in memory 
+        if len(detox_config.main.exclude_if_on) > 0:
+            for replica in all_replicas:
+                ds_name = replica.dataset.name
+                for sitename in detox_config.main.exclude_if_on:
+                    if self.inventory_manager.store.check_if_on(ds_name,sitename):
+                        replica.dataset.demand['on_protected_site'] = True
+
         logger.info('Saving site and dataset states.')
 
         # update site and dataset lists

lib/detox/variables.py

@@ -104,6 +104,16 @@ def _get(self, dataset):
         except KeyError:
             return 0.
 
+class DatasetOnProtectedSite(DatasetAttr):
+    def __init__(self):
+        DatasetAttr.__init__(self, Attr.BOOL_TYPE)
+
+    def _get(self, dataset):
+        try:
+            return dataset.demand['on_protected_site']
+        except KeyError:
+            return False
+
 class ReplicaSize(DatasetReplicaAttr):
     def __init__(self):
         DatasetReplicaAttr.__init__(self, Attr.NUMERIC_TYPE)
@@ -272,6 +282,7 @@ def _get(self, site):
     'dataset.usage_rank': DatasetUsageRank(),
     'dataset.demand_rank': DatasetDemandRank(),
     'dataset.release': DatasetRelease(),
+    'dataset.on_protected_site': DatasetOnProtectedSite(),
     'replica.is_last_transfer_source': ReplicaIsLastSource(),
     'replica.size': ReplicaSize(),
     'replica.incomplete': ReplicaIncomplete(),