[
https://issues.apache.org/jira/browse/TIKA-3746?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tim Allison resolved TIKA-3746.
-------------------------------
Fix Version/s: 1.28.3
Resolution: Fixed
I copied/pasted and modified the unit tests in 2.x to 1.x. One of the
challenges with 1.x is that the forked process is not quite as responsive as in
2.x. In 2.x, the forked process has a thread that reads from the stdout of
the forking process, and if that connection goes down, the forked process
terminates immediately(ish). In 1.x, the forked process reads from the shared
tmp/status file and/or checks the forking process's stdout on a "ping" schedule.
The updated unit tests rely on forking processes instead of threads and set a
very fast ping rate for the forked process (50 ms).
> Update 1.x TikaServerIntegrationTest to use processes instead of threads per
> test
> ---------------------------------------------------------------------------------
>
> Key: TIKA-3746
> URL: https://issues.apache.org/jira/browse/TIKA-3746
> Project: Tika
> Issue Type: Wish
> Reporter: Tim Allison
> Priority: Major
> Fix For: 1.28.3
>
>
> Get rid of flaky tests in 1.x.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
