On (10/12/14 10:42), Pavel Březina wrote: >On 12/10/2014 09:18 AM, Lukas Slebodnik wrote: >>On (09/12/14 15:37), Pavel Březina wrote: >>>On 12/08/2014 09:24 PM, Jakub Hrozek wrote: >>>>On Mon, Dec 08, 2014 at 07:45:37PM +0100, Pavel Březina wrote: >>>>>https://fedorahosted.org/sssd/ticket/2521 >>>> >>>>It's much better with this patch than without -- without the patch, I've >>>>seen maybe 30% failure rate, now I see maybe 10%. But I can still >>>>reproduce failures quite easily. >>>> >>>>Would you like to access to my machine or maybe logs? >>> >>>I can no longer reproduce it on your machine with this patch. >>> >>>For some reason the time() differs from gettimeofday() timestamp in one >>>second on the failed tests. I did not dig into the glibc code to see >>>the differences. I suspect there is some microseconds imprecision. >>> >>Yes, it fixxed the issue. >> >>All green tests: >>http://sssd-ci.duckdns.org/logs/job/5/06/summary.html >>http://sssd-ci.duckdns.org/logs/job/5/05/summary.html >>http://sssd-ci.duckdns.org/logs/job/5/04/summary.html >>http://sssd-ci.duckdns.org/logs/job/5/03/summary.html >>http://sssd-ci.duckdns.org/logs/job/5/02/summary.html >>http://sssd-ci.duckdns.org/logs/job/5/01/summary.html >>http://sssd-ci.duckdns.org/logs/job/4/99/summary.html >>http://sssd-ci.duckdns.org/logs/job/4/98/summary.html >>http://sssd-ci.duckdns.org/logs/job/4/97/summary.html >> >>There is a warning >> >> CC src/responder/common/ifp_tests-negcache.o >>src/tests/cmocka/test_be_ptask.c:67:15: error: function declaration isn’t a >>prototype [-Werror=strict-prototypes] >> static time_t get_current_time() >> ^ >>cc1: all warnings being treated as errors >> >>Simple change will fix it. >> -static time_t get_current_time() >> +static time_t get_current_time(void) > >Thank you for the review. New patch is attached. >
>From 2c3d8cf61bd8da99b03ead8a8e31b80903a4b7e4 Mon Sep 17 00:00:00 2001 >From: =?UTF-8?q?Pavel=20B=C5=99ezina?= <pbrez...@redhat.com> >Date: Mon, 8 Dec 2014 19:43:41 +0100 >Subject: [PATCH] be_ptask: use gettimeofday() instead of time() > >Sometimes the timestamp from gettimeofday() slightly differs >from the one obtained via time() which caused unit test to fail on >occasionaly. > >Resolves: >https://fedorahosted.org/sssd/ticket/2521 >--- ACK LS _______________________________________________ sssd-devel mailing list sssd-devel@lists.fedorahosted.org https://lists.fedorahosted.org/mailman/listinfo/sssd-devel
