Matthew Clarke created NIFI-6901:
------------------------------------
Summary: Version Control version change completes even when
cancelled.
Key: NIFI-6901
URL: https://issues.apache.org/jira/browse/NIFI-6901
Project: Apache NiFi
Issue Type: Bug
Components: Core Framework
Affects Versions: 1.9.0
Reporter: Matthew Clarke
The UI presents misleading/incorrect information when user cancels a version
change on a version controlled Process group.
Simple test to reproduce.
Create a test Process group and build below flow inside it:
GenerateFlowFile ---> ExecuteStreamCommand --> UpdateAttribute
Configure the ExecuteStreamCommand processor with configured Command Path set
to "sleep" and configured Command Argument of 300.
Version control this test PG.
edit the ExecuteStreamCommand configured Command Argument to 600.
Now commit version 2 of the test PG to version control.
Start the dataflow.
Now try to "change version" and select version 1.
UI will display "Stopping Affected Components" and then "Stopping Affected
Processors".
It will not progress past this because the ExecuteStreamCommand still has an
active thread. After 10 minutes when thread competes, version change will
complete.
Now let us assume a thread that will never complete (hung) or a thread that may
run for hours. We can click on the "cancel" button in the UI. NiFi displays
that the version change operation has been cancelled.
Looking at the UI we see processor in a stopped state (cancel simply leaves you
at state it was at in the version control process at the time user hit
"cancel").
On top of that, it really did not cancel. If the user right clicks on
ExecuteStreamCommand and selects "terminate", the version change finishes and
starts the processor again.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
