On 04/05/2010 01:56 PM, Sidney Markowitz wrote:
> Am I right in thinking that this has to be a problem with the
> build/test server and not the result of any of the SpamAssassin
> checkins? Does anyone know how to fix this?

That's my thinking, though it looks like it might be resolved as of
build #5250.  The first failure, build #5248, only had Justin's
automated sought update (which should be fine due to the automation):
http://hudson.zones.apache.org/hudson/job/SpamAssassin-trunk/5248/

>From #5248's console output, I see this:
> FATAL: command execution failed
> hudson.util.IOException2: Failed to join the process
...
> FATAL: Unable to delete script file
> /var/tmp/hudson6691967404758152626.sh
> hudson.util.IOException2:
> remote file operation failed: /var/tmp/hudson6691967404758152626.sh
> at hudson.remoting.chan...@3eff14:hudson-solaris (Solaris)
...
> [locks-and-latches] Releasing all the locks
> [locks-and-latches] All the locks released
> FATAL: channel is already closed
> hudson.remoting.ChannelClosedException: channel is already closed

This differs from #5249, which dies MUCH sooner:
> At revision 930883
> [locks-and-latches] Checking to see if we really have the locks
> [locks-and-latches] Have all the locks, build can start
> [locks-and-latches] Releasing all the locks
> [locks-and-latches] All the locks released
> FATAL: Timer already cancelled.
> java.lang.IllegalStateException: Timer already cancelled.

These java errors disappear for build #5250, which actually completes
the build and test:
> All tests successful.
> Files=166, Tests=2052, 159 wallclock secs ( 0.67 usr  0.60 sys
>                               + 40.82 cusr  8.77 csys = 50.86 CPU)
> Result: PASS
> + mv Mail-SpamAssassin-3.4.0/t/log artifacts/t.log.make_disttest 
> + perl ./build/hudson/tap-to-junit-xml make test testxml/make_test
> + perl ./build/hudson/tap-to-junit-xml make disttest
>                                             testxml/make_disttest
> [locks-and-latches] Releasing all the locks
> [locks-and-latches] All the locks released
> Archiving artifacts
> Recording test results
> Sending e-mails to: [email protected] [email protected]
>     [email protected] [email protected] [email protected]
>     [email protected]
> Finished: UNSTABLE

The unstable result comes from a failed meta test due to a dependency
error caused by John Hardin's removal of ADVANCE_FEE_2_NEW as noted at
http://hudson.zones.apache.org/hudson/job/SpamAssassin-trunk/5250/testReport/make_test/t_basic_meta_t/test______2/


My own test run had one more failure coming from dnsbl_subtests:

> Test Summary Report
> -------------------
> t/basic_meta.t                  (Wstat: 0 Tests: 2 Failed: 1)
>   Failed test:  2
> t/dnsbl_subtests.t              (Wstat: 0 Tests: 47 Failed: 30)
>   Failed tests:  2-9, 17-25, 31-43
> Files=169, Tests=2123, 527 wallclock secs ( 1.37 usr  0.28 sys
>                          + 259.42 cusr 16.66 csys = 277.73 CPU)
> Result: FAIL
> Failed 2/169 test programs. 31/2123 subtests failed.
> make: *** [test_dynamic] Error 255


The scrollback (which I've logged if anybody wants it) notes missing
test rules prefixed with X_URIBL_Y*.

Reply via email to