try this one now... :)
Jason Titus
jazzmantitus at yahoo.com
Fri Sep 8 06:08:12 UTC 2006
Yup. Looks like it worked fine. It was a confusing error though.
jason-titus-computer:~/Desktop/memcached-1.2.x-svn372 jasontitus$
perl t/flush-all.t
1..11
[warn] kqueue.c: detected broken kqueue; not using.: No buffer space
available
ok 1 - stored foo
ok 2 - foo == 'fooval'
ok 3 - did flush_all
ok 4 - foo == <undef>
ok 5 # skip flush_all is still only second-granularity. need atomic
counter on flush_all.
ok 6 # skip flush_all is still only second-granularity. need atomic
counter on flush_all.
ok 7 - foo == <undef>
ok 8 - did flush_all in future
ok 9 - stored foo = '1234'
ok 10 - foo == '1234'
ok 11 - foo == <undef>
Keep up the good work,
Jason
On Sep 7, 2006, at 9:23 PM, Brad Fitzpatrick wrote:
> Can you run t/flush-all.t by hand and reply with that output?
>
> There aren't any other errors that I see... just libevent
> complaining that
> Apple still hasn't fixed their kqueue so it's going to use something
> slower.
>
> - Brad
>
> On Thu, 7 Sep 2006, Jason Titus wrote:
>
>> 64-bit Mac OS X 10.4.7 on Intel fails some tests when built w/ 64-bit
>> support ('-m64')
>>
>> Check it out -
>>
>> jason-titus-computer:~/Desktop/memcached-1.2.x-svn372 jasontitus$
>> prove
>> t
>> t/00-startup.........[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/00-startup.........ok
>> t/64bit..............[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/64bit..............ok
>> t/binary-get.........skipped
>> all skipped: Tests not written.
>> t/bogus-commands.....[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/bogus-commands.....ok
>> t/daemonize..........ok
>> t/delete-window......[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/delete-window......ok 14/20# waiting 5 seconds for the deleter
>> event...
>> t/delete-window......ok
>> t/expirations........[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/expirations........ok
>> t/flags..............[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/flags..............ok
>> t/flush-all..........[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/flush-all..........ok 7/11
>> # Failed test 'foo == <undef>'
>> # in /Users/jasontitus/Desktop/memcached-1.2.x-svn372/t/lib/
>> MemcachedTest.pm at line 46.
>> # got: 'VALUE foo 0 4
>> t/flush-all..........NOK 11# 1234
>> # END
>> # '
>> # expected: 'END
>> # '
>> # Looks like you failed 1 test of 11.
>> t/flush-all..........dubious
>> Test returned status 1 (wstat 256, 0x100)
>> DIED. FAILED test 11
>> Failed 1/11 tests, 90.91% okay (less 2 skipped tests: 8
>> okay,
>> 72.73%)
>> t/getset.............[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/getset.............ok
>> t/incrdecr...........[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/incrdecr...........ok
>> t/lru................skipped
>> all skipped: Tests not written.
>> t/managed-buckets....skipped
>> all skipped: Tests not written.
>> t/multiversioning....[warn] kqueue.c: detected broken kqueue; not
>> using.: No buffer space available
>> t/multiversioning....ok
>> t/slab-reassign......skipped
>> all skipped: Tests not written.
>> t/stats..............skipped
>> all skipped: Tests not written.
>> t/udp................skipped
>> all skipped: Tests not written.
>> t/unixsocket.........skipped
>> all skipped: Tests not written.
>> Failed Test Stat Wstat Total Fail List of Failed
>>
>> Jason
>>
>> On Sep 4, 2006, at 6:45 PM, Brad Fitzpatrick wrote:
>>
>>> FreeBSD / Solaris users,
>>>
>>> Could you try this one?
>>>
>>> http://danga.com/memcached/dist/experimental/memcached-1.2.x-
>>> svn362.tar.gz
>>>
>>> It now builds quietly on OS X and passes all tests:
>>>
>>> $ prove test
>>> test/00-startup.........ok
>>> test/64bit..............skipped
>>> all skipped: Skipping 64-bit tests on 32-bit build
>>> test/binary-get.........skipped
>>> all skipped: Tests not written.
>>> test/bogus-commands.....ok
>>> test/daemonize..........ok
>>> test/delete-window......ok 14/20# waiting 5 seconds for the deleter
>>> event...
>>> test/delete-window......ok
>>> test/expirations........ok
>>> test/flags..............ok
>>> test/flush-all..........ok
>>> 2/11 skipped: flush_all is still only second-granularity.
>>> need atomic counter on flush_all.
>>> test/getset.............ok
>>> test/incrdecr...........ok
>>> test/lru................skipped
>>> all skipped: Tests not written.
>>> test/managed-buckets....skipped
>>> all skipped: Tests not written.
>>> test/multiversioning....ok
>>> test/slab-reassign......skipped
>>> all skipped: Tests not written.
>>> test/stats..............skipped
>>> all skipped: Tests not written.
>>> test/udp................skipped
>>> all skipped: Tests not written.
>>> test/unixsocket.........skipped
>>> all skipped: Tests not written.
>>> All tests successful, 8 tests and 2 subtests skipped.
>>> Files=18, Tests=94, 31 wallclock secs (11.30 cusr + 4.08 csys =
>>> 15.38
>>> CPU)
>>>
>>>
>>> So I imagine it was the same issue (below) for FreeBSD, etc.
>>> Let me
>>> know.
>>>
>>> - Brad
>>>
>>>
>>> ---------- Forwarded message ----------
>>> Date: Tue, 5 Sep 2006 01:41:43 +0000 (UTC)
>>> From: commits at code.sixapart.com
>>> To: memcached-commits at lists.danga.com, cvs-commits at livejournal.com
>>> Subject: [memcached] bradfitz,
>>> r362: the old unixsocket check didn't work on ...
>>>
>>> the old unixsocket check didn't work on OS X or other BSDs or
>>> Solaris... making it more explicit.
>>>
>>> U trunk/server/memcached.c
>>>
>>>
>>> Modified: trunk/server/memcached.c
>>> ===================================================================
>>> --- trunk/server/memcached.c 2006-09-05 00:22:48 UTC (rev 361)
>>> +++ trunk/server/memcached.c 2006-09-05 01:41:43 UTC (rev 362)
>>> @@ -1296,7 +1296,7 @@
>>> /* unix socket mode doesn't need this, so zeroed out. but
>>> why
>>> * is this done for every command? presumably for UDP
>>> * mode. */
>>> - if (c->request_addr.sa_family != AF_UNSPEC) {
>>> + if (!settings.socketpath) {
>>> c->request_addr_size = sizeof(c->request_addr);
>>> } else {
>>> c->request_addr_size = 0;
>>>
>>>
>> --jasontitus
>>
>>
More information about the memcached
mailing list