NIH | National Cancer Institute | NCI Wiki  

Date

Attendees



Goals

Discussion items

TimeItemWhoNotes

Java 17
  • Migration to Java 17
    • Terms browser jenkins/Tomcat working
      • Issues with cart forwarding to http address - fixed now with URL repoint
      • Excel export failing - possible POI conflict - fixed now with URL repoint
        • What is the use case? Can we drop it
        • Large value sets are the issue and csv works.
        • Have pointer to FTP site?
        • examine use of this feature
    • CTS2 build/Jenkins working.
      • Tomcat has longstanding issues - VM tracker running
    • NCIm work started
    • NCI Term Form - Check for Java 17 tag. Might be fixed, needs testing
      • Need Jenkins path for new tomcat
  • Will work on services one at a time on Dev and QA
    • Will send to new Stage and Prod
    • Will cutover when all finished
  • Some of the value set files seem to want to live on local machine - compiledVSDefinitions?
      • Do code review to determine where dependencies lie - create Jira item


Tomcat 9 is a priority, might need to sever from Java 17.


CTS2
  • Chasing ehcache temp bean jar creation
    • Normal behavior, but all files should be deleted when application shuts down/restarted
    • Claims files cant be used or processed - loading files fails. 
    • Kicks off at same time as commonValueSetsUtils. Does a hash of all value sets
    • ehcache updated as part of Spring 5 update
    • Heap Dump file might be useful during startup to see what is being loaded onto stack
      • onchange() no longer hooked up for managing process, since no longer loading during serving
      • Could be adding a lot of pressure on environment to build it all during startup
      • Could new OS be handling the stacks differently?
      • Corretto - outside chance this is an issue, but we think it not likely



Production NCI Terms
  • Still multiple issues on Prod
    • Scott creating Jira issues as items identified
    • Possible do daily restart of NCITerms.  Ticket entered - but resistance met.
      • The long restart time could alarm the on-call tech
  • L3cache issues?  Possibly filling up and not allowing processing by cores
    • Dev has larger cache and more CPUs
  • Possible security solution - throttle IPs that are overworking CPU
    • L5 load balancer
  • Mapping versions are null 
  • Check code to see where versions are being pulled

They have initiated tracking on this server. F5 Analytics has been enabled, will monitor to see what kind of data is produced.


System migration
  • Prod and Stage completed Testing
  • DataQA after Prod
    • database on machine - migrate first?
    • Test performance of old loaders vs new database.
    • Other apps follow DB - determine if needs to be co-located
    • Does script need to be updated?
  • List of EVS URLs
  • Will install tomcat 9 and Java 17
  • Check with systems - See if we can test prod2 data deployment
  • Start process for migrating DataQA to new server - Tracy

SumoLogic to DataDog
  • All logs reporting
  • Working on Dashboards and reports
  • Testing refresh from archive
  • Attempting to import logs from Jan 1, 2024 onward

Tier changes

Dev and QA not sync'd for 6 months. Will look at updating this month after monthlies

  • Tracy to put in ticket

DataQA old JVM

Brandon Devarra scanned DataQA and found it on Oracle Java 8.  Has it drifted from Prod?

-- 189356 (1) - OpenJDK 8 <= 8u392 / 11.0.0 <= 11.0.21 / 17.0.0 <= 17.0.9 / 21.0.0 <= 21.0.1 Multiple Vulnerabilities (2024-01-16


Path : /usr/local/jdk1.8/

Installed version : 8.0.391

Fixed version : Upgrade to a version greater than 8u392

Full DataQA needs to be migrated to new environment.  


  • $PATH on devvm points to Oracle jdk.  Should be Corretto
  • Going to reboot to update OS - might impact drive mounts
  • This may all be moot with the Java 17 / new hardware

Processing server
  • Check on JVM - is 8u411
  • What is OS on this machine

Protege servers

Dev and QA now of Corretto

  • Have systems check upper tiers - need ticket

Is Protege prod on a scheduled restart?


FTP

Having issues with S3 bucket

Might be impacting displayed versions of file mappings

Action items

  •