DIRAC 9.0.9__tar.gz → 9.0.10__tar.gz

{dirac-9.0.9/src/DIRAC.egg-info → dirac-9.0.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: DIRAC
-Version: 9.0.9
+Version: 9.0.10
 Summary: DIRAC is an interware, meaning a software framework for distributed computing.
 Home-page: https://github.com/DIRACGrid/DIRAC/
 License: GPL-3.0-only
@@ -19,7 +19,7 @@ Requires-Dist: cachetools
 Requires-Dist: certifi
 Requires-Dist: cwltool
 Requires-Dist: diraccfg
-Requires-Dist: DIRACCommon==v9.0.9
+Requires-Dist: DIRACCommon==v9.0.10
 Requires-Dist: diracx-client>=v0.0.1
 Requires-Dist: diracx-core>=v0.0.1
 Requires-Dist: diracx-cli>=v0.0.1

{dirac-9.0.9 → dirac-9.0.10}/release.notes

@@ -1,3 +1,17 @@
+[v9.0.10]
+
+*DataManagementSystem
+
+FIX: (#8381) make sure the DataManager knows the VO
+
+*Interfaces
+
+FIX: (#8380) DConfigCache - reset Registry caches after loading the configuration from the local disk cache
+
+*TransformationSystem
+
+CHANGE: (#8373) This contains various speed of for the TS, mostly by using temporary tables for the queries, only requesting what's needed in the DB, and bulking things more efficiently
+
 [v9.0.9]
 
 CHANGE: (#8359) Subprocess: use psutil for killing processes

{dirac-9.0.9 → dirac-9.0.10}/setup.cfg

@@ -26,7 +26,7 @@ install_requires =
 	certifi
 	cwltool
 	diraccfg
-	DIRACCommon==v9.0.9
+	DIRACCommon==v9.0.10
 	diracx-client >=v0.0.1
 	diracx-core >=v0.0.1
 	diracx-cli >=v0.0.1

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/Core/Utilities/ElasticSearchDB.py

@@ -82,7 +82,6 @@ def generateDocs(data, withTimeStamp=True):
 
 
 class ElasticSearchDB:
-
     """
     .. class:: ElasticSearchDB
 
@@ -506,7 +505,7 @@ class ElasticSearchDB:
             indexName = self.generateFullIndexName(indexPrefix, period)
         else:
             indexName = indexPrefix
-        sLog.debug(f"Bulk indexing into {indexName} of {data}")
+        sLog.debug(f"Bulk indexing into {indexName} of {len(data)}")
 
         res = self.existingIndex(indexName)
         if not res["OK"]:

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/DataManagementSystem/Client/DataManager.py

@@ -10,11 +10,11 @@ This module consists of DataManager and related classes.
 """
 
 # # imports
-from datetime import datetime, timedelta
+import errno
 import fnmatch
 import os
 import time
-import errno
+from datetime import datetime, timedelta
 
 # # from DIRAC
 import DIRAC
@@ -25,13 +25,13 @@ from DIRAC.Core.Utilities.File import makeGuid, getSize
 from DIRAC.Core.Utilities.List import randomize, breakListIntoChunks
 from DIRAC.Core.Utilities.ReturnValues import returnSingleResult
 from DIRAC.Core.Security.ProxyInfo import getProxyInfo
+from DIRAC.Core.Security.ProxyInfo import getVOfromProxyGroup
 from DIRAC.ConfigurationSystem.Client.Helpers.Operations import Operations
 from DIRAC.DataManagementSystem.Client import MAX_FILENAME_LENGTH
-from DIRAC.MonitoringSystem.Client.DataOperationSender import DataOperationSender
 from DIRAC.DataManagementSystem.Utilities.DMSHelpers import DMSHelpers
+from DIRAC.MonitoringSystem.Client.DataOperationSender import DataOperationSender
 from DIRAC.Resources.Catalog.FileCatalog import FileCatalog
 from DIRAC.Resources.Storage.StorageElement import StorageElement
-from DIRAC.ResourceStatusSystem.Client.ResourceStatus import ResourceStatus
 
 
 # # RSCID
@@ -89,7 +89,7 @@ class DataManager:
         :param vo: the VO for which the DataManager is created, get VO from the current proxy if not specified
         """
         self.log = gLogger.getSubLogger(self.__class__.__name__)
-        self.voName = vo
+        self.voName = vo if vo else getVOfromProxyGroup().get("Value", None)
 
         if catalogs is None:
             catalogs = []
@@ -97,10 +97,9 @@ class DataManager:
 
         self.fileCatalog = FileCatalog(catalogs=catalogsToUse, vo=self.voName)
         self.accountingClient = None
-        self.resourceStatus = ResourceStatus()
         self.ignoreMissingInFC = Operations(vo=self.voName).getValue("DataManagement/IgnoreMissingInFC", False)
         self.useCatalogPFN = Operations(vo=self.voName).getValue("DataManagement/UseCatalogPFN", True)
-        self.dmsHelper = DMSHelpers(vo=vo)
+        self.dmsHelper = DMSHelpers(vo=self.voName)
         self.registrationProtocol = self.dmsHelper.getRegistrationProtocols()
         self.thirdPartyProtocols = self.dmsHelper.getThirdPartyProtocols()
         self.dataOpSender = DataOperationSender()

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/Interfaces/Utilities/DConfigCache.py

@@ -9,6 +9,7 @@ from DIRAC import gLogger
 from DIRAC.Core.Base.Script import Script
 from DIRAC.Core.Utilities.File import secureOpenForWrite
 from DIRAC.ConfigurationSystem.Client.ConfigurationData import gConfigurationData
+from DIRAC.ConfigurationSystem.Client.Helpers.Registry import reset_all_caches
 
 
 class ConfigCache:
@@ -69,5 +70,6 @@ class ConfigCache:
             try:
                 with open(self.configCacheName, "rb") as fh:
                     gConfigurationData.mergedCFG = pickle.load(fh)
+                    reset_all_caches()
             except:
                 gLogger.error("Cache corrupt or unreadable")

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/TransformationSystem/Agent/InputDataAgent.py

@@ -74,7 +74,10 @@ class InputDataAgent(AgentModule):
         """Main execution method"""
 
         # Get all the transformations
-        result = self.transClient.getTransformations({"Status": "Active", "Type": self.transformationTypes})
+        result = self.transClient.getTransformations(
+            {"Status": "Active", "Type": self.transformationTypes},
+            columns=["TransformationID", "AuthorDN", "AuthorGroup"],
+        )
         if not result["OK"]:
             self.log.error("InputDataAgent.execute: Failed to get transformations.", result["Message"])
             return S_OK()

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/TransformationSystem/Agent/MCExtensionAgent.py

@@ -1,4 +1,4 @@
-""" Agent to extend the number of tasks given the Transformation definition
+"""Agent to extend the number of tasks given the Transformation definition
 
 The following options can be set for the MCExtensionAgent.
 
@@ -8,6 +8,7 @@ The following options can be set for the MCExtensionAgent.
   :dedent: 2
   :caption: MCExtensionAgent options
 """
+
 from DIRAC import S_OK, gLogger
 from DIRAC.Core.Base.AgentModule import AgentModule
 from DIRAC.ConfigurationSystem.Client.Helpers.Operations import Operations
@@ -54,7 +55,9 @@ class MCExtensionAgent(AgentModule):
             return S_OK("Disabled via CS flag")
 
         # Obtain the transformations in Cleaning status and remove any mention of the jobs/files
-        res = self.transClient.getTransformations({"Status": "Active", "Type": self.transformationTypes})
+        res = self.transClient.getTransformations(
+            {"Status": "Active", "Type": self.transformationTypes}, columns=["TransformationID", "MaxNumberOfTasks"]
+        )
         if res["OK"]:
             for transDict in res["Value"]:
                 transID = transDict["TransformationID"]

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/TransformationSystem/Agent/TaskManagerAgentBase.py

@@ -231,7 +231,9 @@ class TaskManagerAgentBase(AgentModule, TransformationAgentsUtilities):
             selectCond["Type"] = transType
         if agentType:
             selectCond["AgentType"] = agentType
-        res = self.transClient.getTransformations(condDict=selectCond)
+        res = self.transClient.getTransformations(
+            condDict=selectCond, columns=["TransformationID", "Body", "Author", "AuthorGroup"]
+        )
         if not res["OK"]:
             self.log.error("Failed to get transformations:", res["Message"])
         elif not res["Value"]:

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/TransformationSystem/Agent/TransformationCleaningAgent.py

@@ -144,7 +144,8 @@ class TransformationCleaningAgent(AgentModule):
 
         # Obtain the transformations in Cleaning status and remove any mention of the jobs/files
         res = self.transClient.getTransformations(
-            {"Status": TransformationStatus.CLEANING, "Type": self.transformationTypes}
+            {"Status": TransformationStatus.CLEANING, "Type": self.transformationTypes},
+            columns=["TransformationID", "Author", "AuthorGroup", "Type"],
         )
         if res["OK"]:
             for transDict in res["Value"]:
@@ -161,7 +162,10 @@ class TransformationCleaningAgent(AgentModule):
             self.log.error("Failed to get transformations", res["Message"])
 
         # Obtain the transformations in RemovingFiles status and removes the output files
-        res = self.transClient.getTransformations({"Status": "RemovingFiles", "Type": self.transformationTypes})
+        res = self.transClient.getTransformations(
+            {"Status": "RemovingFiles", "Type": self.transformationTypes},
+            columns=["TransformationID", "Author", "AuthorGroup"],
+        )
         if res["OK"]:
             for transDict in res["Value"]:
                 if self.shifterProxy:
@@ -183,6 +187,7 @@ class TransformationCleaningAgent(AgentModule):
             {"Status": TransformationStatus.COMPLETED, "Type": self.transformationTypes},
             older=olderThanTime,
             timeStamp="LastUpdate",
+            columns=["TransformationID", "Author", "AuthorGroup"],
         )
         if res["OK"]:
             for transDict in res["Value"]:
@@ -230,7 +235,10 @@ class TransformationCleaningAgent(AgentModule):
             return res
         transformationIDs = res["Value"]
         if transformationIDs:
-            res = self.transClient.getTransformations({"TransformationID": transformationIDs})
+            res = self.transClient.getTransformations(
+                {"TransformationID": transformationIDs},
+                columns=["TransformationID", "Status", "Author", "AuthorGroup", "Type"],
+            )
             if not res["OK"]:
                 self.log.error("Failed to get transformations", res["Message"])
                 return res

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/TransformationSystem/Agent/ValidateOutputDataAgent.py

@@ -76,7 +76,9 @@ class ValidateOutputDataAgent(AgentModule):
         self.updateWaitingIntegrity()
         gLogger.info("-" * 40)
 
-        res = self.transClient.getTransformations({"Status": "ValidatingOutput", "Type": self.transformationTypes})
+        res = self.transClient.getTransformations(
+            {"Status": "ValidatingOutput", "Type": self.transformationTypes}, columns=["TransformationID"]
+        )
         if not res["OK"]:
             gLogger.error("Failed to get ValidatingOutput transformations", res["Message"])
             return res
@@ -98,7 +100,7 @@ class ValidateOutputDataAgent(AgentModule):
     def updateWaitingIntegrity(self):
         """Get 'WaitingIntegrity' transformations, update to 'ValidatedOutput'"""
         gLogger.info("Looking for transformations in the WaitingIntegrity status to update")
-        res = self.transClient.getTransformations({"Status": "WaitingIntegrity"})
+        res = self.transClient.getTransformations({"Status": "WaitingIntegrity"}, columns=["TransformationID"])
         if not res["OK"]:
             gLogger.error("Failed to get WaitingIntegrity transformations", res["Message"])
             return res

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/TransformationSystem/Client/TransformationClient.py

@@ -114,13 +114,21 @@ class TransformationClient(Client):
         newer=None,
         timeStamp=None,
         orderAttribute=None,
-        limit=100,
+        limit=None,
         extraParams=False,
         columns=None,
     ):
         """gets all the transformations in the system, incrementally. "limit" here is just used to determine the offset."""
         rpcClient = self._getRPC()
 
+        # If the body is requested (or is served by default)
+        # we take smaller chunk, not to take too much memory
+        # on the server
+        if columns and "Body" not in columns:
+            limit = 100_000
+        else:
+            limit = 1_000
+
         transformations = []
         if condDict is None:
             condDict = {}

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC/TransformationSystem/DB/TransformationDB.py

@@ -6,6 +6,9 @@ This class is typically used as a base class for more specific data processing
 databases
 """
 
+# Disable it because pylint does not understand decorator (convertToReturnValue)
+
+# pylint: disable=invalid-sequence-index
 import re
 import time
 import threading
@@ -15,6 +18,7 @@ from errno import ENOENT
 from DIRAC import gLogger, S_OK, S_ERROR
 from DIRAC.Core.Base.DB import DB
 from DIRAC.Core.Utilities.DErrno import cmpError
+from DIRAC.Core.Utilities.ReturnValues import convertToReturnValue, returnValueOrRaise
 from DIRAC.Resources.Catalog.FileCatalog import FileCatalog
 from DIRAC.Core.Security.ProxyInfo import getProxyInfo
 from DIRAC.Core.Utilities.List import stringListToString, intListToString, breakListIntoChunks
@@ -25,6 +29,7 @@ from DIRAC.DataManagementSystem.Client.MetaQuery import MetaQuery
 
 MAX_ERROR_COUNT = 10
 
+TMP_TABLE_JOIN_LIMIT = 100
 #############################################################################
 
 
@@ -270,6 +275,7 @@ class TransformationDB(DB):
         self.__updateTransformationLogging(transID, message, author, connection=connection)
         return S_OK(transID)
 
+    @convertToReturnValue
     def getTransformations(
         self,
         condDict=None,
@@ -289,32 +295,54 @@ class TransformationDB(DB):
             columns = self.TRANSPARAMS
         else:
             columns = [c for c in columns if c in self.TRANSPARAMS]
-        req = "SELECT {} FROM Transformations {}".format(
-            intListToString(columns),
-            self.buildCondition(condDict, older, newer, timeStamp, orderAttribute, limit, offset=offset),
-        )
-        res = self._query(req, conn=connection)
-        if not res["OK"]:
-            return res
-        if condDict is None:
-            condDict = {}
-        webList = []
+
+        join_query = ""
+
+        try:
+            # If we request multiple TransformationIDs, and they are more than TMP_TABLE_JOIN_LIMIT,
+            # we create a temporary table to speed up the query
+            if (
+                "TransformationID" in condDict
+                and isinstance(condDict["TransformationID"], list)
+                and len(condDict["TransformationID"]) > TMP_TABLE_JOIN_LIMIT
+            ):
+                # Create temporary table for TransformationIDs
+                transIDs = condDict.pop("TransformationID")
+                sqlCmd = "CREATE TEMPORARY TABLE to_query_TransformationIDs (TransID INTEGER NOT NULL, PRIMARY KEY (TransID)) ENGINE=MEMORY;"
+                returnValueOrRaise(self._update(sqlCmd, conn=connection))
+                join_query = " JOIN to_query_TransformationIDs t ON TransformationID = t.TransID"
+
+                # Insert TransformationIDs into temporary table
+                sqlCmd = "INSERT INTO to_query_TransformationIDs (TransID) VALUES ( %s )"
+                returnValueOrRaise(self._updatemany(sqlCmd, [(transID,) for transID in transIDs], conn=connection))
+
+            req = "SELECT {} FROM Transformations {} {}".format(
+                intListToString(columns),
+                join_query,
+                self.buildCondition(condDict, older, newer, timeStamp, orderAttribute, limit, offset=offset),
+            )
+            matching_transformations = returnValueOrRaise(self._query(req, conn=connection))
+
+        finally:
+            # Clean up temporary table
+            if join_query:
+                sqlCmd = "DROP TEMPORARY TABLE to_query_TransformationIDs"
+                self._update(sqlCmd, conn=connection)
+
+        # TODO: optimize by getting all the extra params at once
         resultList = []
-        for row in res["Value"]:
+        for row in matching_transformations:
             # Prepare the structure for the web
-            rList = [str(item) if not isinstance(item, int) else item for item in row]
             transDict = dict(zip(columns, row))
-            webList.append(rList)
             if extraParams:
-                res = self.__getAdditionalParameters(transDict["TransformationID"], connection=connection)
-                if not res["OK"]:
-                    return res
-                transDict.update(res["Value"])
+                trans_extra_param = returnValueOrRaise(
+                    self.__getAdditionalParameters(transDict["TransformationID"], connection=connection)
+                )
+
+                transDict.update(trans_extra_param)
             resultList.append(transDict)
-        result = S_OK(resultList)
-        result["Records"] = webList
-        result["ParameterNames"] = columns
-        return result
+
+        return resultList
 
     def getTransformation(self, transName, extraParams=False, connection=False):
         """Get Transformation definition and parameters of Transformation identified by TransformationID"""
@@ -710,21 +738,38 @@ class TransformationDB(DB):
         countDict["Total"] = sum(countDict.values())
         return S_OK(countDict)
 
+    @convertToReturnValue
     def __addFilesToTransformation(self, transID, fileIDs, connection=False):
-        req = "SELECT FileID from TransformationFiles"
-        req = req + " WHERE TransformationID = %d AND FileID IN (%s);" % (transID, intListToString(fileIDs))
-        res = self._query(req, conn=connection)
-        if not res["OK"]:
-            return res
-        for tupleIn in res["Value"]:
-            fileIDs.remove(tupleIn[0])
-        if not fileIDs:
-            return S_OK([])
-        values = [(transID, fileID) for fileID in fileIDs]
-        req = "INSERT INTO TransformationFiles (TransformationID,FileID,LastUpdate,InsertedTime) VALUES (%s, %s, UTC_TIMESTAMP(), UTC_TIMESTAMP())"
-        if not (res := self._updatemany(req, values, conn=connection))["OK"]:
-            return res
-        return S_OK(fileIDs)
+        # Create temporary table for FileIDs
+        sqlCmd = "CREATE TEMPORARY TABLE to_query_FileIDs (FileID INT(11) UNSIGNED NOT NULL, PRIMARY KEY (FileID)) ENGINE=MEMORY;"
+        returnValueOrRaise(self._update(sqlCmd, conn=connection))
+
+        try:
+            # Insert FileIDs into temporary table
+            sqlCmd = "INSERT INTO to_query_FileIDs (FileID) VALUES ( %s )"
+            returnValueOrRaise(self._updatemany(sqlCmd, [(fileID,) for fileID in fileIDs], conn=connection))
+
+            # Query existing files using JOIN
+            req = (
+                "SELECT tf.FileID FROM TransformationFiles tf JOIN to_query_FileIDs t ON tf.FileID = t.FileID WHERE tf.TransformationID = %d;"
+                % transID
+            )
+            res = returnValueOrRaise(self._query(req, conn=connection))
+
+            # Remove already existing fileIDs using set difference for efficiency
+            existingFileIDs = {tupleIn[0] for tupleIn in res}
+            fileIDs = list(set(fileIDs) - existingFileIDs)
+            if not fileIDs:
+                return []
+
+            values = [(transID, fileID) for fileID in fileIDs]
+            req = "INSERT INTO TransformationFiles (TransformationID,FileID,LastUpdate,InsertedTime) VALUES (%s, %s, UTC_TIMESTAMP(), UTC_TIMESTAMP())"
+            returnValueOrRaise(self._updatemany(req, values, conn=connection))
+            return fileIDs
+        finally:
+            # Clean up temporary table
+            sqlCmd = "DROP TEMPORARY TABLE to_query_FileIDs"
+            returnValueOrRaise(self._update(sqlCmd, conn=connection))
 
     def __insertExistingTransformationFiles(self, transID, fileTuplesList, connection=False):
         """Inserting already transformation files in TransformationFiles table (e.g. for deriving transformations)"""
@@ -1271,18 +1316,35 @@ class TransformationDB(DB):
     # These methods manipulate the DataFiles table
     #
 
+    @convertToReturnValue
     def __getFileIDsForLfns(self, lfns, connection=False):
         """Get file IDs for the given list of lfns
         warning: if the file is not present, we'll see no errors
         """
-        req = f"SELECT LFN,FileID FROM DataFiles WHERE LFN in ({stringListToString(lfns)});"
-        res = self._query(req, conn=connection)
-        if not res["OK"]:
-            return res
-        lfns = dict(res["Value"])
-        # Reverse dictionary
-        fids = {fileID: lfn for lfn, fileID in lfns.items()}
-        return S_OK((fids, lfns))
+
+        if not lfns:
+            return ({}, {})
+        # Create temporary table for LFNs
+        sqlCmd = "CREATE TEMPORARY TABLE to_query_LFNs (LFN VARCHAR(255) NOT NULL, PRIMARY KEY (LFN)) ENGINE=MEMORY;"
+        returnValueOrRaise(self._update(sqlCmd, conn=connection))
+
+        try:
+            # Insert LFNs into temporary table
+            sqlCmd = "INSERT INTO to_query_LFNs (LFN) VALUES ( %s )"
+            returnValueOrRaise(self._updatemany(sqlCmd, [(lfn,) for lfn in lfns], conn=connection))
+
+            # Query using JOIN with temporary table
+            req = "SELECT df.LFN, df.FileID FROM DataFiles df JOIN to_query_LFNs t ON df.LFN = t.LFN;"
+            res = returnValueOrRaise(self._query(req, conn=connection))
+
+            lfns = dict(res)
+            # Reverse dictionary
+            fids = {fileID: lfn for lfn, fileID in lfns.items()}
+            return (fids, lfns)
+        finally:
+            # Clean up temporary table
+            sqlCmd = "DROP TEMPORARY TABLE to_query_LFNs"
+            self._update(sqlCmd, conn=connection)
 
     def __getLfnsForFileIDs(self, fileIDs, connection=False):
         """Get lfns for the given list of fileIDs"""

{dirac-9.0.9 → dirac-9.0.10/src/DIRAC.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: DIRAC
-Version: 9.0.9
+Version: 9.0.10
 Summary: DIRAC is an interware, meaning a software framework for distributed computing.
 Home-page: https://github.com/DIRACGrid/DIRAC/
 License: GPL-3.0-only
@@ -19,7 +19,7 @@ Requires-Dist: cachetools
 Requires-Dist: certifi
 Requires-Dist: cwltool
 Requires-Dist: diraccfg
-Requires-Dist: DIRACCommon==v9.0.9
+Requires-Dist: DIRACCommon==v9.0.10
 Requires-Dist: diracx-client>=v0.0.1
 Requires-Dist: diracx-core>=v0.0.1
 Requires-Dist: diracx-cli>=v0.0.1

{dirac-9.0.9 → dirac-9.0.10}/src/DIRAC.egg-info/requires.txt

@@ -4,7 +4,7 @@ cachetools
 certifi
 cwltool
 diraccfg
-DIRACCommon==v9.0.9
+DIRACCommon==v9.0.10
 diracx-client>=v0.0.1
 diracx-core>=v0.0.1
 diracx-cli>=v0.0.1