Angelo Fausti added a comment - 10/Mar/21 7:33 PM Accomplishments in this epic: The EFD replication service is running on the stable cluster at NCSA and continuously replicating data from the Summit EFD. SQR-050 shows the result of the investigation in which we selected MirrorMaker 2 as the replacement for the Confluent Replicator (37%). We deployed MirrorMaker 2 and the JDBC sink connectors on the int cluster at NCSA and let them running as a proof of concept to show that we can continuously replicate a topic from the Summit EFD and write to the Consolidated Database at NCSA. The next step is to define the schema for the EFD data we need in the Consolidated database, and use perhaps Kafka KSQL to combine multiple Kafka topics into a single table in postgres (15%). We implemented a solution to handle NaN values in the EFD and showed in SQR-053 that the symbol NaN can be used to represent missing values for floating-point data types across all systems involved in the data flow from the CSCs to the EFD client (20%). We practiced and documented data migration from the current SQuaSH production instance to a SQuaSH sandbox instance deployed with phalanx. We didn't get to replace the SQuaSH production instance during this cycle though (17%). Support to SQuaSH and EFD instances (21%).