You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@usergrid.apache.org by gr...@apache.org on 2015/12/12 20:52:37 UTC
usergrid git commit: Added comments explaining tool.
Repository: usergrid
Updated Branches:
refs/heads/unique_index_logging 9e526ba5b -> 9ebbc1b90
Added comments explaining tool.
Project: http://git-wip-us.apache.org/repos/asf/usergrid/repo
Commit: http://git-wip-us.apache.org/repos/asf/usergrid/commit/9ebbc1b9
Tree: http://git-wip-us.apache.org/repos/asf/usergrid/tree/9ebbc1b9
Diff: http://git-wip-us.apache.org/repos/asf/usergrid/diff/9ebbc1b9
Branch: refs/heads/unique_index_logging
Commit: 9ebbc1b903c74d36d775d2581b72795c5ac85797
Parents: 9e526ba
Author: George Reyes <gr...@apache.org>
Authored: Sat Dec 12 11:52:35 2015 -0800
Committer: George Reyes <gr...@apache.org>
Committed: Sat Dec 12 11:52:35 2015 -0800
----------------------------------------------------------------------
.../tools/ManagementUserIndexMissingFix.java | 36 ++++++++++++++------
1 file changed, 25 insertions(+), 11 deletions(-)
----------------------------------------------------------------------
http://git-wip-us.apache.org/repos/asf/usergrid/blob/9ebbc1b9/stack/tools/src/main/java/org/apache/usergrid/tools/ManagementUserIndexMissingFix.java
----------------------------------------------------------------------
diff --git a/stack/tools/src/main/java/org/apache/usergrid/tools/ManagementUserIndexMissingFix.java b/stack/tools/src/main/java/org/apache/usergrid/tools/ManagementUserIndexMissingFix.java
index 2a6b00f..5d075fc 100644
--- a/stack/tools/src/main/java/org/apache/usergrid/tools/ManagementUserIndexMissingFix.java
+++ b/stack/tools/src/main/java/org/apache/usergrid/tools/ManagementUserIndexMissingFix.java
@@ -75,20 +75,26 @@ import static org.apache.usergrid.persistence.cassandra.Serializers.ue;
/**
- * This utility audits all values in the ENTITY_UNIQUE column family. If it finds any duplicates of users then it
- * deletes the non existing columns from the row. If there are no more columns in the row then it deletes the row. If
- * there exists more than one existing column then the one with the most recent timestamp wins and the other is
- * deleted.
+ * This utility audits all values in the ENTITY_UNIQUE column family. The way it does this is by taking in a *.json file
+ * that has an array of users then checking them to see if they are in the unique column family. Logging as it goes along.
*
- * If you want the run the tool on their cluster the following is what you need to do nohup java
- * -Dlog4j.configuration=file:log4j.properties -jar usergrid-tools-1.0.2.jar UserUniqueIndexCleanup -host
- * <cassandra_host_here> > log.txt
+ * An additional use of this tool is to use the -dup flag like so
+ * java -Dlog4j.configuration=file:log4j.properties -jar usergrid-tools-1.0.2.jar ManagementUserAudit -host localhost -dup grey@apache.org
*
- * if there is a specific value you want to run the tool on then you need the following
+ * If we find a value that is present and also exists in ENTITY_PROPERTIES then it goes and queries the entity index.
+ * The reason this check has to be done is that it can be possible for an entity to have an invalid uuid in the entity index
+ * but has the correct data in ENTITY_UNIQUE. Causing a user to be unable to login.
+ * If they are in the index then we see if the uuid's that are stored in the index correspond to any entity in the ENTITY_PROPERTIES
+ * column family. If it does not then we retreive the column name from the RAW query then delete that column name from the bucket
+ * that contains the data. Thus freeing up the uuid to be queried by user name.
*
- * nohup java -Dlog4j.configuration=file:log4j.properties -jar usergrid-tools-1.0.2.jar UserUniqueIndexCleanup -host
- * <cassandra_host_here> -app <applicationUUID> -col <collection_name> -property <unique_property_key> -value
- * <unique_property_value> > log.txt
+ * THIS FIX SHOULD ONLY BE APPLIED IF PROBLEMS MATCH THESE OTHER CONDITIONS:
+ *
+ * When running the audit tools if you see two different uuid's for the same entity AND you cannot log in with email.
+ * If the unique audit found that this email should be working but the entity_index audit found it to be broken.
+ * IF you can PUT on the entity_unique uuid to update properties but CANNOT put on entity_index uuid.
+ *
+ * The above three conditions should be satisfied before this is ran on ANY user.
*
* @author grey
*/
@@ -261,18 +267,26 @@ public class ManagementUserIndexMissingFix extends ToolBase {
logger.info( "Trying to remove uuid: {} from ENTITY_INDEX.", r.getRef().getUuid() );
+ //This method didn't exist before and requires exposing the buffer with the ScanColumn information
+ //in the query processor. Hacky, but better than other methods.
List<ScanColumn> entityIds = relationManager.searchRawCollection( "users", query );
for ( ScanColumn entityId : entityIds ) {
SecondaryIndexSliceParser.SecondaryIndexColumn secondaryIndexColumn =
( SecondaryIndexSliceParser.SecondaryIndexColumn ) entityId;
+ //byte buffer from the scan column is the column name needed to delete from ENTITY_INDEX
DynamicComposite columnName = dce.fromByteBuffer( secondaryIndexColumn.getByteBuffer() );
String bucketId =
( ( EntityManagerImpl ) em ).getIndexBucketLocator().getBucket( r.getRef().getUuid() );
Object index_name = key( MANAGEMENT_APPLICATION_ID, "users", "email" );
+ //only delete from entity_index. Other data might contain valid references which is why
+ //preconditions should be satisfied before trying this method. Preconditions satisfy that
+ //uuid doesn't have valid data anywhere else but in INDEX for specific email. May still have the
+ //broken uuid else where but not of concern to users.
+
Object index_key = key( index_name, bucketId );
logger.info( "Deleting the following rowkey: {} from ENTITY_INDEX.", index_key );
addDeleteToMutator( m, ENTITY_INDEX, index_key, columnName, createTimestamp() );