DX Operational Intelligence

doi 1.3.2 installation problems

  • 1.  doi 1.3.2 installation problems

    Posted 10-25-2019 11:09 AM
    Hello all,

    I am trying to install doi v.1.3.2 in a disconnected network where we have our own registry server with the doi images loaded. It is not 100% successful, and as I am new to both Openshift and DOI, I am struggling to find out what is wrong. Hope you can help!

    The installation log has the following error message:

    Error response from from daemon: Get https://doi.packages.ca.com/v1/users/: dial tcp: lookup doi.packages.ca.com on xxx.xxx.xxx.xxx:53: server misbehaving
    Warning:: Docker login failed.

    Later in the installation log, the DOI_DOCKER_REGISTRY parameter is correctly set to our local registry, so it seems that at one place, it is ignoring this and wants to access Broadcom's registry instead.

    The rest of the installation finishes, and most of pods are deployed OK, with the exceptions of

    clientdatanode1: Stays on rolling deployment forever
    jobmanager: Stays on rolling deployment forever
    namenode: goes into a Crash Loop Back-Off state. The log says: java.io.IOException: Cannot create directory /opt/ca/hadoop/dfs/name/current
    resourcemanager1: Stays on rolling deployment forever
    resourcemanager2: Stays on rolling deployment forever

    And there are several deployment configs with zero pods running: acn-correlation, doi-kibana, elastalert, filebeatmonitoring, logstashmonitring, kube-state-metrics and clientnodemanager.

    Any ideas? Could the cause be that the installation wants to pull images from Broadcoms registry, rather than our own?

    Best regards
    Olav