2016-04-17 18:23 GMT+02:00 Simon Slavin <slavins at bigfraud.org>: > > On 17 Apr 2016, at 5:13pm, Cecil Westerhof <cldwesterhof at gmail.com> wrote: > > > Or is this normal > > in Java programs? > > Yes. You're discovering that Java is rubbish. 'futex' for Java is what > most environments call 'mutex' and Java doesn't do locking well. It's this > which is causing your delays, not SQLite or anything around it. > > Google 'java futex' if you want details, but there's little point since > you can't do much about it. Just stay updated to the latest OS and the > latest Java available to you and hope someone figured it out. >
?I put a strace on it. This was what I got: Process 26455 attached with 20 threads % time seconds usecs/call calls errors syscall ------ ----------- ----------- --------- --------- ---------------- 99.80 11245.498406 42527 264435 130887 futex 0.09 10.480000 3493333 3 fsync 0.08 8.886784 0 39275508 read 0.02 2.552284 0 16397440 write 0.00 0.367716 0 50708676 lseek 0.00 0.048000 8000 6 6 restart_syscall 0.00 0.000442 1 737 sched_yield 0.00 0.000000 0 2 close 0.00 0.000000 0 1 stat 0.00 0.000000 0 4 fstat 0.00 0.000000 0 4 mmap 0.00 0.000000 0 128 mprotect 0.00 0.000000 0 6 rt_sigprocmask 0.00 0.000000 0 1 rt_sigreturn 0.00 0.000000 0 2 2 access 0.00 0.000000 0 4 madvise 0.00 0.000000 0 7 fcntl 0.00 0.000000 0 3 unlink 0.00 0.000000 0 1 gettid 0.00 0.000000 0 2 sched_getaffinity ------ ----------- ----------- --------- --------- ---------------- 100.00 11267.833632 106646970 130895 total ? ?Pretty damaging for futex: almost half of the calls went wrong and 99.8% of the time is spend there?. ?I have some finding out to do.? -- Cecil Westerhof
