Issues fixed in MarkLogic Server 12.0.0
- Last Updated: April 6, 2026
- 4 minute read
- MarkLogic Server
- Documentation
Released: July 29, 2025
The following issues have been addressed in this release:
| Bug ID | Details |
|---|---|
| MLE-22316 | Automatic cache sizing on machines with NUMA CPU architecture and 376GB or more of RAM will result in an invalid number of compressed tree cache partitions. This can cause the Admin UI to throw a 500 error with the following message: XDMP-LEXVAL: gr:memory-partitions("33") -- Invalid lexical value "33" with namespace '[http://marklogic.com/xdmp/group|http://marklogic.com/xdmp/group]' using schema 'group.xsd' |
| MLE-22284 | Certain SPARQL queries that use inference within an Optic query that also does a join can result in a segmentation fault. |
| MLE-21731 | There is a rounding error in the swap size calculation when running MarkLogic AMIs on AWS. This can lead to invalid warnings in the error log indicating that the recommended swap space is not configured. |
| MLE-21728 | When using the Amazon Linux 2023 AMIs, the secondary network interface is not detected correctly. |
| MLE-21507 | The /etc/MarkLogic/proc-util.sh script in 11.3.1 for RHEL 9 uses a hard coded MARKLOGIC_USER and MARKLOGIC_INSTALL_DIR values. These should be read from the /etc/marklogic.conf file. |
| MLE-21504 | When using the admin.uiSetBanner() function to configure a banner for the server UIs, line break tags are displayed in the message rather than processed correctly. |
| MLE-21458 | Backups and other operations against AWS S3 in large clusters may be very slow and result in timeouts. |
| MLE-21414 | xdmp:zip-manifest() with a zero-byte file causes a segmentation fault when it should return an XDMP-INVZIP error. |
| MLE-21078 | Coupling clusters when running 11.3.1 may result in an error due to the presence of the The workaround is to remove this property from the payload when posting the payload to |
| MLE-20665 | The Optic joinDoc() function does not automatically cast the second argument to the required type. This results in the function not returning the expected results if the argument is not the correct type. |
| MLE-20182 | mlcp import fails with an XDMP-DOCSTARTTAGCHAR error when importing XML aggregates where a default namespace (no prefix is defined) is used. This only happens when running on Java 11 or later. |
| MLE-19886 | In the Admin UI cluster status page, if there are large binary files, the reported database size is different than the size reported on the database status page because the size reported on the database status page includes the size of the large binary data as well. |
| MLE-19860 | The xdmp:database-describe-indexes() function returns an incorrect value for the parentLocalname for Range Element Attribute Index entries. |
| MLE-19853 | There is host IP caching at the XDQP protocol layer that prevents MarkLogic database replication from working across networks that use NAT and there is no mechanism to disable this caching. |
| MLE-19270 | Returning JSON results in Query Console that have certain encoded strings in the property values (e.g. "\r\n: 7e5"), will result in an error "Parser Error: Cannot parse result as JSON". |
| MLE-19201 | Starting in 11.3.1, the Monitoring History charts are slow to render and sometimes get out of memory errors when loading metrics for a lot of databases or spanning a large timeframe. |
| MLE-18255 | Under heavy request load, the reported app server task queue may go beyond the max task queue size when the task queue is full. |
| MLE-18111 | Under certain circumstances, there may be a long delay in forest failover due to other hosts in the cluster trying to run the "detect indexes" process against hosts that are not available. |
| MLE-16678 | When deleting a database that has forests that are attached, the attached forests are detached automatically but the message 'No database with identifier' will appear in the error log even though the database is deleted successfully. |
| MLE-16433 | When under heavy load, incremental backup may not correctly create the “catalog” of documents and then on a subsequent restore, the XDMP-BACKUPCATALOGMISSING error will be logged. |
| MLE-16029 | There is a startup issue with MarkLogic Server version 11.3 when installed on AWS EC2 instances running RHEL 8.10 or 8.9. The server fails to start due to a missing Java JVM, which is required for EC2 integration. This issue does not occur in MarkLogic Server versions 11.1 and 11.2. Workaround: Create a configuration file |
| MLE-15922 | Under certain circumstances, no error is thrown when a large binary document write fails. Instead, the only indication of the failure is that the following log message appears when MarkLogic is shut down 'Warning: BinaryManager::decrementBinaryRefCount'. |
| MLE-14273 | SPARQL queries with optional bind parameters may cause unnecessary locking in the in-memory stand when the system is processing large ingestion or update workloads. |
| MLE-13867 | xdmp:document-filter() fails on certain Word documents. This is due to a bug in the underlying library that is used to extract text from the binary document formats. |
| MLE-12132 | Exporting Optic pipelines that include cts.geospatialRegionPathReference() queries results in an error: JS-JAVASCRIPT: fn.string(container.pathExpression), -- Error running JavaScript request: TypeError: Cannot read property 'pathExpression' of undefined. |
| MLE-834 | Path range indexes of type string that select a JSON array node directly (instead of the values in the array) do not put values in the range index because the array cannot be converted into a string. This silently fails without any messages in the logs even if the invalid values setting is set to Workaround: Do not select the array node directly but select the values of the array. |
| MLE-773 | The PDF conversion process may throw an XDMP-ATTRSEQ error when creating XML from XHTML that contains empty anchors tags. |
| MLE-735 | When importing a workspace in Query Console, the query tabs that were open when the workspace was exported will not be automatically opened after importing. |
| MLE-534 | SPARQL query performance of SELECT * WHERE {?s ?p ?o} LIMIT BY 10 can be dramatically different depending upon the constraining collections and whether the results are ordered by the subject, predicate or object. |
| MLE-450 | Under rare conditions, a segmentation fault occurs in the BIO_set() call during a LDAPS external security request. This is due to an issue in the underlying OpenSSL library. |
| MLE-403 | A TDE template that contains a sub-template that has the exact columns and column settings as the parent template is an invalid template but does not fail validation. The template will insert successfully but it will not behave correctly. |