Rocket DevOps (formerly Aldon)

 View Only

Performance considerations on startup

  • 1.  Performance considerations on startup

    Posted 09-29-2022 10:12
    Edited by Dale Asher 09-29-2022 10:11

    We recently heard from one of our Rocket DevOps customers on IBMi. They were seeing well over 5 minutes passing after startup before they were able to log in to RDOp. So they were wondering:

    Are there any further performance tuning tips or recommendations available ?

    We found that they had plenty of memory (12G RAM) and sufficient processing on their server which was dedicated to hosting RDOp and RDOi.  

    After local review of newer system options, we identified the configuration item that was causing the bottleneck for java processing on startup.

    Below is the excerpt now included in our installation guide for notes on the issue and the steps to resolve.


    Jiqing Fan (Rico)
    Rocket Internal - All Brands

    Rocket DevOps portal - Performance considerations on startup

    Startup performance of Rocket DevOps on IBM i could be affected by the sharing mode on your logical partition.

    LPARs with shared processors can have a sharing mode of CAPPED (the LPAR can access only the processing power of its assigned units) or UNCAPPED (the LPAR can access additional processors in the shared pool).

    To see the LPAR settings, run this command:


    Reviewing the output from the command, in the PARTITION INFO section, you will see either SHARING MODE: UNCAPPED or SHARING MODE: CAPPED.

    CAPPED sharing mode could cause Rocket DevOps to take longer for startup (up to 10 minutes on some systems), but should not affect Rocket DevOps performance once it is running.

    UNCAPPED sharing mode ensures sharing of processor capacity during startup and will likely result in improved performance (faster application deployment and startup).

    However, changing one LPAR to UNCAPPED could affect the workload of other LPARs. Consult with your IT department about making any changes to the sharing mode.