SR-A8236 · Issue 217418
Removed default overriding threads per node setting of Topology page
Resolved in Pega Version 7.2
The number of threads per node specified in Decision->Simulation Settings->Topology landing page was not being taken into account by the data flow execution, leading to the inability to configure vertical scalability per PRPC node and/or exclude certain nodes from data flow execution. This was traced to a default value overriding the setting from this page, and the code has been updated to correctly reflect the setting being used.
SR-A8099 · Issue 216453
Null/empty values removed from propagation instead of being set to 0
Resolved in Pega Version 7.2
Numeric properties without a value set were being propagated as value 0 during data flow execution. This was due to a clipboard page error in the Data Flow batch run execution, and this has now been updated to to ensure null or empty properties will not be propagated in the data flow.
SR-A10599 · Issue 221392
Privilege handling added for PAD split-scheme database setup
Resolved in Pega Version 7.2
in the "Predictive Analytics Director Settings" landing page, designating the internal Database to an external database rule caused the landing page to hang and an error was logged. This was caused by a lack of privileges for the Database user associated with PAD. PAD needs privileges to create, update, delete, select and insert. In cases where the user is associated with no privileges, PAD failed to create tables in the first instance. if the config was changed, PAD tried to check if any open projects were present and would fail again because of privileges and lack of tables. To correct this, additional handling has been added: A user who has insufficient privileges will generate an exception and the user will not be able to make any PAD operations. A user switching from an insufficient privileged user to a privileged user will need to re-login. Newly granting privileges to a insufficient privileged user and relaunching the portal will not allow user to create projects: the server will need to restarted to take the changes into account.
SR-A9936 · Issue 222962
Handling added to check for Snappy library for DNodeBootstrap
Resolved in Pega Version 7.2
The initialization of DNodeBootstrap at PRPC startup was generating an error if Snappy (a native library for Linux OS and ppc64 arch) was not present. To resolve this, handling has been added to check whether the Snappy library can be initialized.
SR-A13188 · Issue 228542
Apache HTTP client library updated for improved security
Resolved in Pega Version 7.2
The Apache HTTP client library version 4.3.5 was identified as containing a security vulnerability. Both DSM and PRPC have been updated to version 4.3.6 in the DSM & PRPC SVN repositories respectively. In addition, new ADM service archives (WAR/EAR) need to be created with the updated library. VBD (service) does not use this library, so is not affected.
SR-A14475 · Issue 228318
Local ADM 'Fail Safe' settings honored in PostgreSQL
Resolved in Pega Version 7.2
When the ADM to was configured to persist learnings, the learnings/training data of ADM models was lost when server is restarted. This was traced to the default FailSafeEnabled setting of TRUE in PostgreSQL overriding the disabling of the ?FAIL SAFE? mode option from Decisioning -> Infrastructure -> Services. This mode puts the JDBC connection away from auto-commit and fails to put it back to auto-commit, thereby making ADM backups not commit to the database. This was on a per connection basis, so some connections would still save, but others would not. This has been fixed by explicitly ensuring all connections have the required auto-commit settings on ADM Server connections.
SR-A13888 · Issue 227100
D-nodes automated for sequential startup
Resolved in Pega Version 7.2
D-Nodes require that servers are restarted sequentially one after each other. In installations with a large cluster, this restart may get out of sync and result in some nodes not joining the cluster properly because of being restarted together. To improve this, the D-Node bootstrap procedure has been automated for an orderly restart to ensure no two D-nodes start in parallel.
SR-A13157 · Issue 225924
Performance enhancements in DSM
Resolved in Pega Version 7.2
Performance enhancements have been made to DSM. These include improving the timing info for Switch, Champion Challenger, Exclusion, and Data Join, and adding a Smart clone feature of SR pages if a check indicates that cloning would minimize memory footprint.
SR-A13157 · Issue 224896
Performance enhancements in DSM
Resolved in Pega Version 7.2
Performance enhancements have been made to DSM. These include improving the timing info for Switch, Champion Challenger, Exclusion, and Data Join, and adding a Smart clone feature of SR pages if a check indicates that cloning would minimize memory footprint.
SR-A10575 · Issue 220506
DSM monitoring and troubleshooting enhancements
Resolved in Pega Version 7.2
Monitoring and troubleshooting improvements have been made to the DSM components D-Node, VBD and ADM. These include an HTTP call which returns a node status that can be integrated into monitoring tools and raise appropriate alerts, and services to provide diagnostics information about d-nodes that can be sent by the cloud team to engineers for troubleshooting.