-
Notifications
You must be signed in to change notification settings - Fork 17
Changelog: 7.x.x Releases
Doreen Seider edited this page Dec 1, 2015
·
14 revisions
- Fixed uploading of files and directories to remote RCE instances (users that run RCE 7.0.0 in a distributed RCE network should upgrade to 7.0.1 as soon as possible)
- Fixed publishing of workflows (to make them available via remote access)
- Fixed determining the optimal design values in the Optimizer in case 'Vector' as data type is used
- New workflow components: Database, Evaluation Memory
- Remote tool access via SSH connections (secure way to execute tools even if an untrusted network exists in between)
- Improved network communication system (scalability, generated system load)
- Advances in fault-tolerant and (multi-)nested loops
- Various bugfixes and improvements
- Reworked structure of example workflows to function as a kind of tutorial
- Enhanced Workflow Execution Wizard
- Added ability to assign RCE instances to multiple selected workflow components in one step
- Added feature to assign any remote RCE instance randomly
- Added ability to group the table with the workflow components by its type
- Improved filter options
- Added full-text search for the table with the workflow components
- Prevent from closing wizard immediately after opened in highly distributed environments
- Reviewed and improved log messages of all workflow components
- Reworked nested-loop capability so that multi-nested loops are possible (introduced more restrict rules which are explained in the user guide and also in the workflow examples)
- Extended the capability of fault-tolerant loops: each driver workflow component (Optimizer, Design of Experiments, Converger, Parametric Study) is able to decide about the fault-tolerance of the loop driven
- Output Writer supports simple data types (values of type float, integer, short text, etc. can be written in a file with a user-defined format)
- Optimizer writes results in a csv file and stores it in the data management
- Added new workflow component 'Evaluation Memory' which is able to store evaluation results for later re-use
- Design of Experiments, Converger, and Parametric Study are now nested-loop capable
- Added workflow component "Database" which supports select, insert, update, and delete statements for MySQL and PostgreSQL databases
- Made cancelling of Optimizer more robust
- Set SQL Reader, SQL Writer and SQL Command to deprecated
- Eliminated deprecated CPACS tool wrapper
- Cluster component supports relative paths
- Converger: fixed incrementing number of convergence checks for multiple inputs
- Workflow driver components (Optimizer, Design of Experiments, Converger, Parametric Study) are able to forward values
- Extended validation for workflow components: error is shown if an input is not connected that is declared as required
- Improved routing so that labels do not interfere with connection lines anymore
- Made the textual representation of workflow files more steady on workflow modifications
- Made parsing of workflow files more robust against modified integrated tools (workflow is still opened, but the affected tool is left out)
- Log messages of workflow components are completely available in the workflow data browser
- Error log messages are aggregated for a workflow in one single log file which is made available in the workflow data browser
- The workflow file for each executed workflow is available in the workflow data browser
- XPaths of dynamic inputs/outputs of XML and CPACS workflow components are shown in the workflow data browser
- Changed default for storing workflow data (formerly: history data): they are stored by default for all workflow components now (can still be disabled)
- Added support for user documentation (one single PDF or *.txt file; max 50 MB)
- Added a so called "Tool run imitation mode": kind of dummy behavior can be defined for integrated tools that imitates the actual tool run (useful to test workflow logic or debug certain tools in a loop)
- Copying of integrated tools is always done sequentially (to avoid I/O load peaks at workflow start if many tools of a workflow need to be copied)
- Default working directory of tools is not the RCE's temp directory anymore (as this results in using the RCE's temp directory even if the working directories are used for inspection during and after workflow execution)
- Default handling and default execution constraints can be defined for inputs by the tool integrator when integrating the tool
- Added command "wf delete" that deletes a certain workflow from the data management
- Improved log output: added additional relevant information, and removed irrelevant messages
- Added correlation markers ("E#...") to connect observed errors on one machine to root causes that happened on another Linux .deb packages do not strictly require the OpenJDK JRE anymore (although this is still the tested and recommended option)
- Added a startup option to run a headless instance without local components (--disable-components)
- Improved scalability of the network communication system, and reduced the system load generated by it. This increases the maximum possible number of clients per relay, increases message throughput, and improves stability when there is high external CPU load on the system (e.g. from running integrated tools on the same machine).
- Reduced network traffic by reducing and combining individual messages.
- Improved performance of the database used to store all of the workflow data.
- Passwords for SSH accounts are no longer configured as plain text, but as a non-recoverable ("hashed") format instead. Plain-text configuration still works, but is discouraged; a warning will be logged on startup.
- As the new password format cannot be generated manually, a special configuration interface for SSH accounts has been added ("rce --configure"). To support editing SSH accounts on servers, this UI can also be run in text mode.
- As a precaution, RCE was hardened against a common network vulnerability. Although RCE is intended to be run in trusted networks only, users are still encouraged to upgrade to RCE 7.x as soon as possible.
- SSH connections to other RCE instances can now be added in the network view
- Tools that are published with the remote access template can now be used by remote instances via SSH connections
- Increased the range of valid remote tool and workflow ids (names) and versions
- Potentially risky characters are now prevented from being used in parameter strings (for example the backslash, or double quotes)
- Fixed that on some Linux distributions clicking on web links in help view or PDF files caused the JVM to crash
- The workflow list view shows workflow state icons
- Various minor bugfixes and improvements
- Improved database update methods, guarded against opening a database that was created with a future version of RCE
- Global lock for XML mapping to avoid memory peaks if big XML files are processed
- Updated TiGL Viewer binaries to version 2.1.6
- Guarded the profile directory against being used with older RCE versions
- Reworked start up validation and pre-workflow execution code to make it more maintainable
- Improved handling of endpoint data of type 'Small Table' (workflow data browser, workflow console, etc.)
- Improvements in workflow engine: robustness, failure handling