Issues fixed in MarkLogic Server 10.0-6

SPARQL Query XDMP-MEMCANCELED

In some circumstances, SPARQL Query is failing with XDMP-MEMCANCELED. For the cases identified, SPARQL query optimizations are improved with this release.

False warning in health view on replica cluster

With correctly configured and functioning master and replica clusters the /manage/v2 health view may give state=at-risk and codes HEALTH-FOREST-FOREIGN-REPLICA-ERROR and HEALTH-DATABASE-FAILED-MASTER-FORESTS, and /v2/databases/Documents?view=status on the replica cluster may show failed-masters.

Documents with both metadata and properties fragments sometimes cannot find the metadata

Sometimes the document-get-metadata and document-get-metadata-value built-in functions return empty sequence for some documents that have properties fragments. After reindexing the problem moves to other documents.

xdmp.jsonValidate() returns an error when validating a valid object against a JSON schema

xdmp.jsonValidate() unexpectedly returns an error when an empty object is validated against an "allOf" JSON schema in certain cases.

Response Headers for HTTPS enabled App-Servers should include strict-transport-security

Added configurable options to control HSTS header in HTTPS response

SPARQL negation operator(!) inside Filter (Not) Exists SegFault

SPARQL Query using negation operator(!) inside Filter Exists/Not Exists results in a segmentation fault

multiple groups for a row set in the Optic API

Previously, the Optic API could aggregate a single group over any row set similar to the SQL GROUP BY statement. Now, the Optic API can aggregate the union of multiple groups over a single row set similar to the SQL GROUPING SET, ROLLUP, and CUBE statements. The Optic API provides a groupByUnion() operation as well as the rollup(), cube(), groupToArrays(), and facetBy() conveniences.

op.in() operator for comparing a value with a list

The Optic API now provides an op.in() operator similar to SQL and SPARQL that compares a test value such as a column with a list of candidates.

Query Console JS profiler retains error printout

The error for JS profiler is sometimes stuck with the old error.

The CPU profile generated by MarkLogic Server is not compatible with Chrome Dev Tools

The CPU profile generated by MarkLogic Server does not display correct execution times when loaded in Chrome Dev Tools' JavaScript Profiler tab

Admin GUI: Configuring database replication using chrome browser fails

Configuring database replication fails with 403 error on newer versions of chrome browser

Query Console: URIs with multiple whitespaces is normalized

URIs with multiple white-spaces gets normalized in the explore list view.

Inaccurate App Server Request Rate Meters

The metered app server and xdbc server request rates are too large. They should match the number of request log entries in the access log.

Query DSL requests for /v1/rows endpoint

Previously, the REST API supported only a JSON AST representation for Optic queries. While this representation is convenient for processors (including the programmatic builders in the Java and Node.js APIs), the JSON AST is not easy for people to read or edit. Thus, in client environments other than the Java and Node.js APIs, Optic queries were difficult to construct. Now, the client can send a representation similar to building an Optic query in SJS but using a strict subset of JavaScript syntax. Because the query is parsed instead of evaluated, the execution strategy doesn't raise security concerns.

Journal path for Backup and Restore missing from Admin UI

Unable to specify Journal Path for Backup or Restore using the Admin UI

Bad HTTP Expect header

An HTTP connection is closed upon an empty "Expect" header. An empty "Expect" header.is an invalid value per the HTTP spec.

SQL, SPARQL SELECT, and QBV XML queries on the /v1/rows endpoint of the REST API

The /v1/rows endpoint of the REST API returns rows matched by queries. Previously, the endpoint only accepted Optic queries in the JSON AST format. Now, the endpoint also accepts queries in the following formats: * SQL queries as application/sql * SPARQL SELECT queries as application/sparql * Query Based View queries as application/xml

JAVA_HOME set in /etc/marklogic.conf is not found

If an AMI has Java installed in a location besides /usr/java/default, /usr/java/latest/, /usr/java/jdk1.[678]*, /usr/lib/jvm/java or /usr/lib/jvm/jre-1.[678].0-*.x86_64x it will not be found even if properly set in /etc/marklogic.conf

cts:element-attribute-range-query throws an error when executing an age query

cts:element-attribute-range-query throws an error when the query is constructed with a value of type xs: xs:yearMonthDuration and the range element attribute index is configured for xs:dateTime

Optic export of path range queries with namespaces

The Optic API can export a plan constrained by a cts.query Previously, if the constraining query included path range query with namespaces, the export didn't serialize the namespaces, so the plan couldn't be imported. Now, the export declares the namespaces.

MarkLogic doesn't recognise custom environment variables

Custom environment variables defined in /etc/marklogic.conf are ignored during MarkLogic start.

CREATE VIEW segfault while using ODBC

The following is broken over ODBC: SQL => create view a.b as select * from customer;

Memory leak using Kerberos with disabled Relay Cache

When using Kerberos with the Relay cache disabled a memory leak will occur. Kerberos library has been upgraded to 1.18.3 in this version of MarkLogic Server.

Query Console mostly frozen

A workspace is open, but no query tab is selected, and none can be selected.

Database bulk Replication Fails with "SVC-FILOPN: File open error"

Stand copy fails in bulk database replication if the copy takes more than 5 minutes, resulting in the replica database unable to sync with the master database.

Empty SAML ID

MarkLogic server does not correctly process SAML authentication when samlId is empty

Telemetry ErrorLog to include Server startup messages

Telemetry .... Stage/Logs/ErrorLog.txt now includes server start messages, sync to systems ErrorLog, for every server re-start with Telemetry log level not Disabled, so to provide enviroment parameters for support purposes.

Tableau connector certificate expired

While trying to connect to the MarkLogic beta connector in Tableau, a notification pops up that the certificate has expired and the connector is unusable

Slow performance issue with SPARQL LIMIT Query

Slow performance issue with SPARQL LIMIT Query using FILTER or VALUES clause

Certificate Authorities appear in Admin UI after deletion

When a Certificate Authority has more than one certificate, and one certificate is deleted, it remains visible in the Admin UI. The list of certificates will only stop displaying once the last certificate for that authority is removed. In the Security database, the certificates still appear, but are disabled, so they will not be used to validate certificates.

cts:triple-range-query() defaults to "uncached"

The cts:triple-range-query() function defaults to "uncached", and the "uncached" option results in "cached" behaviour.

op.fromSearch() / op:from-search() accessor

Adds an Optic function to retrieve the relevance factors for documents matched by a query. In addition, the op.fromSearchDocs() / op:from-search-doc() convenience function provides a convenience for retrieving the documents and uris. Instead of retrieving documents, the relevance factors can be used to filter rows, triples, or range index values by using an inner join on the fragment id for the documents that are the source for the rows, triples, or range index values.

bindAs() / op:bind-as() operation

The Optic API provides an op.as() / op:as() function to add columns. When the op.as() / op:as() function is used in a select() / op:select(), any columns not listed are dropped, making it necessary to list columns to be preserved when adding new columns. That's inconvenient when only adding columns. The bindAs() / op:bind-as() operation adds a new column without dropping any of the existing columns.

Geo Region Index retains index data on disk after deletion

if we have two similar indexes configured(same coordinate systems and precision) with just the geohash precision different, and then if you delete one of those indexes , the data in geohash atlas, locator and other files does not get deleted for the deleted index after a forest merge.

Scheduled backup failure when forests are failed over should be reported as Error msg in ErrorLog

Scheduled backup failure when forests are failed over should be reported as Error msg in ErrorLog

New cts:reference-namespaces() namespaces accessor for path-reference

cts:reference-namespaces() / cts.referenceNamespaces() returns map:map or null, according to path-reference or geospatial-path-reference's results

REST API doesn't clear database

REST API doesn't clear database when forests are failed over

New operation to roll back database forests

Added new operation in REST management api rollback-forests-to-nonblocking-timestamp

MLCP does not work well with DHS autoscaling capability

MLCP uses a fixed number of threads specified by the users in the command line before the job starts running. As the load increases and the DHS cluster scales out, MLCP becomes the bottleneck of the performance if the thread count is low. If the specified thread count is too high at the beginning, MLCP overloads the server.

Query console profiler returns XDMP-CAST error for negative time durations

Query console profiler can sometimes return an XDMP-CAST error with negative times computed for some durations.

Get protected paths not returned

GET /manage/v2/protected-paths does not include protected paths for a user that has the http://marklogic.com/xdmp/privileges/manage privilege but not the manage-admin role.

Wrong error code on /manage/v3 REST endpoint when LDAP user has manage-admin but no security role

MarkLogic /manage/v3 REST endpoint returns HTTP 404 when LDAP mapped user does not have appropriate set of roles. It should be "SEC-DUNOPRIV: Admin error. Default user does not have privileges to this server."