Am 04.05.2016 um 11:42 schrieb Mark Thomas:
Version 1.2.7 includes the following changes:
- Windows binaries rebuild with OpenSSL 1.0.2h
The proposed release artefacts can be found at [1],
and the build was done using tag [2].
The Apache Tomcat Native 1.2.7 is
[X] Stable, go ahead and release
[ ] Broken because of ...
+1 to release, thanks for RMing.
Detailed results especially about unit test failures see below.
There were two JVM crashes, but that's not a regression on my setup. I
did some partial analysis, see at the end of the message.
We might still want to remove some resolved items from TODO.txt and
probably can resolve some more for the next releases.
- Tested with APR 1.5.2 and OpenSSL 1.0.2h
- Platforms Solaris 10 Sparc, SLES 11 64 Bits, RHEL 6 64 Bits
- configure flag "--enable-maintainer-mode"
- Using Java versions 1.8.0_72 for tcnative build, tests/examples
and TC trunk unit tests
- Using "-XX:-UseCompressedClassPointers" in combination with
Java 8 on 64 Bit Linux
- SHA1 and MD5 OK
- signatures OK
- gz and zip for sources consistent
- source dist consistent with svn tag
- config.guess and config.sub from apr 1.5.2 (copied by buildconf)
2 years old, not so nice but still OK.
- VERSIONS says OpenSSL 1.0.2h and APR 1.5.2
- recreated release with jnirelease script, results are
consistent with source dist, except for minor expected diffs in
CHANGELOG.txt and generated docs
- make succeeds and builds lib
- no C warnings
- tcnative tests OK
- unit test results for TC trunk (apr, nio and nio2)
- Tests failing for all connectors
and all platforms:
- org.apache.tomcat.util.net.openssl.ciphers.TestCipher
- org.apache.tomcat.util.net.openssl.ciphers.
TestOpenSSLCipherConfigurationParser
- no further Tests fail on Solaris
- Tests failing on both Linux distributions for all connectors
- org.apache.catalina.core.TestAsyncContextImpl
(all failures in TesterAccessLogValve.validateAccessLog
line 93, so I'll add test.relaxTiming="true" next time).
- Tests failing for RHEL 6
- NIO2:
- org.apache.catalina.tribes.group.TestGroupChannelMemberArrival
Testcase: testMemberArrival took 15.511 sec
FAILED
Checking member arrival length for [Listener-10]. Was [8] but should
have been [9]
Checking member arrival length for [Listener-6]. Was [8] but should have
been [9]
Checking member arrival length for [Listener-5]. Was [8] but should have
been [9]
junit.framework.AssertionFailedError: Checking member arrival length for
[Listener-10]. Was [8] but should have been [9]
Checking member arrival length for [Listener-6]. Was [8] but should have
been [9]
Checking member arrival length for [Listener-5]. Was [8] but should have
been [9]
at
org.apache.catalina.tribes.group.TestGroupChannelMemberArrival.testMemberArrival(TestGroupChannelMemberArrival.java:94)
- org.apache.catalina.tribes.group.interceptors.
TestOrderInterceptor
Testcase: testOrder1 took 17.577 sec
FAILED
null
junit.framework.AssertionFailedError
at
org.apache.catalina.tribes.group.interceptors.TestOrderInterceptor.testOrder1(TestOrderInterceptor.java:96)
- org.apache.tomcat.websocket.TestWsWebSocketContainer
Testcase: testWriteTimeoutClientEndpoint took 12.32 sec
FAILED
Time out was [11836] ms
junit.framework.AssertionFailedError: Time out was [11836] ms
at
org.apache.tomcat.websocket.TestWsWebSocketContainer.doTestWriteTimeoutClient(TestWsWebSocketContainer.java:374)
at
org.apache.tomcat.websocket.TestWsWebSocketContainer.testWriteTimeoutClientEndpoint(TestWsWebSocketContainer.java:309)
- APR:
- org.apache.coyote.http2.TestHttp2Section_5_1
(JVM Crash see below)
- Tests failing on SLES 11:
- APR
- org.apache.catalina.nonblocking.TestNonBlockingAPI
(JVM Crash see below)
- org.apache.tomcat.websocket.TestWsWebSocketContainer
Testcase: testSessionExpirySession took 12.25 sec
FAILED
expected:<1> but was:<0>
junit.framework.AssertionFailedError: expected:<1> but was:<0>
at
org.apache.tomcat.websocket.TestWsWebSocketContainer.testSessionExpirySession(TestWsWebSocketContainer.java:768)
Info about the crashes:
I checked the stacks and both times the crash happened in:
#0 0x0000003f6d232a45 in raise () from /lib64/libc.so.6
#1 0x0000003f6d234225 in abort () from /lib64/libc.so.6
#2 0x0000003f6d22b9d5 in __assert_fail () from /lib64/libc.so.6
#3 0x00007fcde806946a in Java_org_apache_tomcat_jni_Socket_timeoutSet
(e=<value optimized out>, o=<value optimized out>, sock=<value optimized
out>,
timeout=<value optimized out>) at src/network.c:1188
although it happened for different tests and different machines. Both on
64 Bit Linux distributions. There's no Hot Spot error file. The line
thta triggers the assertion is:
1188 TCN_ASSERT(s->opaque != NULL);
The crash on RHEL in org.apache.coyote.http2.TestHttp2Section_5_1
happens at the end of the first unit test after the stopping message,
but before the following line got logged:
org.apache.coyote.AbstractProtocol.destroy Destroying ProtocolHandler
The crash on SLES11 in
org.apache.catalina.nonblocking.TestNonBlockingAPI happens close to the
end of the test, also during shutdown:
[junit] INFO: Stopping service Tomcat
[junit] May 05, 2016 1:01:14 PM org.apache.coyote.AbstractProtocol stop
[junit] INFO: Stopping ProtocolHandler
["http-apr-127.0.0.1-auto-3-16980"]
[junit] May 05, 2016 1:01:14 PM
org.apache.catalina.nonblocking.TestNonBlockingAPI$NBWriteServlet$1
onTimeout
[junit] INFO: onTimeout
[junit] java: src/network.c:1188:
Java_org_apache_tomcat_jni_Socket_timeoutSet: Assertion `(s->opaque !=
((void *)0))' failed.
Regards,
Rainer
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
