Mark, On 9/26/14 3:55 AM, Mark Thomas wrote: > On 25/09/2014 20:56, Christopher Schultz wrote: > >> Some test fail. On my system the "tribes" tests always fail, likely due >> to missing multicast support > > Yes, that will trigger test failures.
I've never explicitly set up multicast, so I've always assumed that it's
not configured. I have multiple Linux environments and an OS X laptop
(where I've never bothered to run the full test-suite).
On Linux, eth0 says:
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
So it seems like multicast should be enabled. Is there special
configuration you know of that might be required in order to run those
tests?
On my Mac, I see multicast as well for both en0 (disconnected ethernet)
and en1 (connected wireless):
en1: flags=8863<UP,BROADCAST,SMART,RUNNING,SIMPLEX,MULTICAST> mtu 1500
Would additional configuration be required to run these tests on the Mac?
>> I noticed this during the test-bio run:
>>
>> [junit] Write length: 1, Buffered: n, Time: 50309480ns
>> [junit] Write length: 1, Buffered: y, Time: 63297451ns
>> [junit] Write length: 10, Buffered: n, Time: 952385ns
>> [junit] Write length: 10, Buffered: y, Time: 1253882ns
>> [junit] Write length: 100, Buffered: n, Time: 643847ns
>> [junit] Write length: 100, Buffered: y, Time: 763425ns
>> [junit] Write length: 1000, Buffered: n, Time: 616628ns
>> [junit] Write length: 1000, Buffered: y, Time: 734361ns
>> [junit] Write length: 10000, Buffered: n, Time: 627050ns
>> [junit] Write length: 10000, Buffered: y, Time: 723313ns
>> [junit] Write length: 100000, Buffered: n, Time: 749909ns
>> [junit] Write length: 100000, Buffered: y, Time: 770355ns
>>
>> Whatever we are buffering, it's not helping ;)
>
> It isn't expected to. Tomcat is buffering internally so the extra buffer
> is just overhead. Ideally, we'd check unbuffered it always faster than
> buffered.
Okay. I was just reading-through the output and noticed a big block of
text that was different from all the test (startup/shutdown,
startup/shutdown, etc.) and it caught my eye.
>> Several tests fail during test-bio, but they all seem to be non-blocking
>> so maybe they aren't supposed to fail:
>>
>> org.apache.catalina.core.ApplicationDispatcher.invoke Servlet.service()
>> for servlet error threw exception
>> [junit] javax.servlet.ServletException: Opps.
>> [junit] at
>> org.apache.catalina.core.TestAsyncContextImpl$ErrorServlet.doGet(TestAsyncContextImpl.java:1028)
>> [junit] at
>> javax.servlet.http.HttpServlet.service(HttpServlet.java:618)
>> [junit] at
>> javax.servlet.http.HttpServlet.service(HttpServlet.java:725)
>>
>> Note that "oops" is misspelled. :)
>
> Those exceptions are all deliberate. As long as the test passes
> everything is fine. I'll see about fixing that typo.
Okay. I double-checked and the "Opps" traces aren't being reported as
test failures or errors. Only the multicast stuff isn't working.
-chris
signature.asc
Description: OpenPGP digital signature
