martinzink commented on a change in pull request #994:
URL: https://github.com/apache/nifi-minifi-cpp/pull/994#discussion_r572079690
##########
File path: nanofi/tests/CSite2SiteTests.cpp
##########
@@ -167,7 +173,9 @@ void different_version_bootstrap(minifi::io::BaseStream*
stream, TransferState&
}
TEST_CASE("TestSiteToBootStrap", "[S2S3]") {
-
+#ifndef WIN32
+ signal(SIGPIPE, SIG_IGN);
Review comment:
Sure, a SIGPIPE is sent to a process if it tried to write to a socket
that had been shutdown for writing or isn't connected (anymore), this can
easily happen in multithreaded enviroment.
The issue is that on macOS the default sigpipe handler crashes the
application.
In production code we are already ignoring this signal by setting the
sigpipe handle to nothing. see
https://github.com/apache/nifi-minifi-cpp/blob/main/main/MiNiFiMain.cpp#L181
Given that in production code we are ignoring the signal, I think it would
be best to ignore it here as well.
I've tested this on my fork with custom CI, on both macos-xcode12.0 and
macos-xcode11.2.1 with 200 runs
`make test ARGS="--timeout 300 -j4 --output-on-failure -R CSite2SiteTests
--repeat-until-fail 200`
Without the fix both configuration failed (both around the 100th run), with
the fix they passed 3 times (that's 1200 runs without failure)
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
