[jira] [Assigned] (GEODE-9462) Dump call stacks from both Dockerized and non-Dockerized java processes

["Dale Emery (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/GEODE-9462?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Dale Emery reassigned GEODE-9462:
---------------------------------

    Assignee: Dale Emery

> Dump call stacks from both Dockerized and non-Dockerized java processes
> -----------------------------------------------------------------------
>
>                 Key: GEODE-9462
>                 URL: https://issues.apache.org/jira/browse/GEODE-9462
>             Project: Geode
>          Issue Type: Improvement
>          Components: build
>            Reporter: Dale Emery
>            Assignee: Dale Emery
>            Priority: Major
>
> Currently, {{ci/scripts/capture-call-stacks.sh}} assumes that if {{PARALLEL_DUNIT}} is empty, tests were run in plain Java processes, and if it is non-empty, tests were run in Docker containers.
> GEODE-8728 violates that assumption: It runs parallel tests in plain Java processes, without Docker containers.
> Currently, the script looks in different places for Java processes, depending on whether {{PARALLEL_DUNIT}} is empty. If it is empty, the script dumps stacks from plain Java processes on the machine, and only those processes. If it is non-empty, the script dumps stacks from Java processes running inside Docker containers, and only those processes.
> This will not work in builds that include GEODE-8728.
> To allow the script to work both for newer builds that include GEODE-8728 and older builds that do not, change it to dump call stacks from both places, regardless of whether `PARALLEL_DUNIT` is empty. If `jps` reports any processes, dump their stacks. And if `docker ps` reports any containers, dump the stacks of the Java processes in each container.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)