[
https://issues.apache.org/jira/browse/MINIFI-188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15848266#comment-15848266
]
marco polo commented on MINIFI-188:
-----------------------------------
Two issues presented yesterday when I debugged this: A malloc failure that is
likely Sierra related ( Mac ). Another was the fact that SIGTERM did not
definitively exit. This is in part because we are attempting to use constructs
within a signal handler that should not be used, namely mutexes and thread
constructs. These are not guaranteed to work as signal handlers are
asynchronous. The mutex could be locked in the start function ( as it was in my
case ), thus creating a deadlock. Switching the construct to a sem_wait and
sem_post is a guaranteed solution and in my case, reliably ends the program.
> Malloc Failure when starting minifi for first time.
> ----------------------------------------------------
>
> Key: MINIFI-188
> URL: https://issues.apache.org/jira/browse/MINIFI-188
> Project: Apache NiFi MiNiFi
> Issue Type: Bug
> Components: C++
> Affects Versions: cpp-0.2.0
> Environment: MacOSX
> Reporter: marco polo
> Assignee: marco polo
> Fix For: cpp-0.2.0
>
>
> I am getting a malloc failure that causes my build to fail the first time.
> This is on my local environment using the vanilla build ( from master ).
> This is preventing me from running my configuration. One of the threads
> begins and fails. The signal handler does not seem to be running either.
> I will look into the issue. This is likely my environment, so I will update
> the ticket as the debugging progresses.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
