Author: remm
Date: Wed Aug 13 14:06:43 2014
New Revision: 1617734
URL: http://svn.apache.org/r1617734
Log:
NIO2 seems less likely to detect a disconnect when writing, for some reason.
When the test fails (extremely rare for me), no error is actually reported by
NIO2 so there's nothing that can really be improved from what I can see. Try to
disconnect earlier to try to improve the test's reliability on gump (this also
makes the test run faster).
Modified:
tomcat/trunk/test/org/apache/catalina/nonblocking/TestNonBlockingAPI.java
Modified:
tomcat/trunk/test/org/apache/catalina/nonblocking/TestNonBlockingAPI.java
URL:
http://svn.apache.org/viewvc/tomcat/trunk/test/org/apache/catalina/nonblocking/TestNonBlockingAPI.java?rev=1617734&r1=1617733&r2=1617734&view=diff
==============================================================================
--- tomcat/trunk/test/org/apache/catalina/nonblocking/TestNonBlockingAPI.java
(original)
+++ tomcat/trunk/test/org/apache/catalina/nonblocking/TestNonBlockingAPI.java
Wed Aug 13 14:06:43 2014
@@ -302,7 +302,7 @@ public class TestNonBlockingAPI extends
int read = 0;
int readSinceLastPause = 0;
int readTotal = 0;
- while (read != -1 && readTotal < WRITE_SIZE / 2) {
+ while (read != -1 && readTotal < WRITE_SIZE / 8) {
long start = System.currentTimeMillis();
read = is.read(buffer);
long end = System.currentTimeMillis();
@@ -349,8 +349,8 @@ public class TestNonBlockingAPI extends
// TODO Figure out why non-blocking writes with the NIO connector
appear
// to be slower on Linux
- alv.validateAccessLog(1, 500, WRITE_PAUSE_MS * 7,
- WRITE_PAUSE_MS * 7 + 30 * 1000);
+ alv.validateAccessLog(1, 500, WRITE_PAUSE_MS,
+ WRITE_PAUSE_MS + 30 * 1000);
}
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
