You are not logged in.
Hi!
This happens sporadically but far too often to ignore. When running java (1.7.0_10, x64, standard kernel) I notice full utilization of all cores; and it's kernel time.
The machine quickly freezes so it's kind of hard to investigate...
$ strace -p 15775
Process 15775 attached
futex(0x7f6e7b8ca9d0, FUTEX_WAIT, 15776, NULL) =
Somebody else having this issue?
thx
tcn
Last edited by tcn (2013-01-02 13:09:50)
Offline
Appears to be fixed with 3.6.11 (?). Plus: code is running much faster now.
Offline
Bummer, just happend again.
Offline
Sounds like a garbage collecting thing. Gc can log its activities. See if the times match your own observations.
Somewhere between "too small" and "too large" lies the size that is just right.
- Scott Hayes
Offline
perf:
65.84% [kernel] [k] _raw_spin_lock_irqsave
strace:
[pid 24947] futex(0x7f52702edd28, FUTEX_WAKE_PRIVATE, 1) = 0
[pid 24947] futex(0x7f52702edd54, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 1, {1357409542, 1448000}, ffffffff) = -1 ETIMEDOUT (Connection timed out)
[pid 24947] futex(0x7f52702edd28, FUTEX_WAKE_PRIVATE, 1) = 0
[pid 24947] futex(0x7f52702edd54, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 1, {1357409542, 11552000}, ffffffff) = -1 ETIMEDOUT (Connection timed out)
[pid 24947] futex(0x7f52702edd28, FUTEX_WAKE_PRIVATE, 1) = 0
[pid 24947] futex(0x7f52702edd54, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 1, {1357409542, 21683000}, ffffffff) = -1 ETIMEDOUT (Connection timed out)
[pid 24947] futex(0x7f52702edd28, FUTEX_WAKE_PRIVATE, 1) = 0
[pid 24947] futex(0x7f52702edd54, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 1, {1357409542, 31788000}, ffffffff) = -1 ETIMEDOUT (Connection timed out)
[pid 24947] futex(0x7f52702edd28, FUTEX_WAKE_PRIVATE, 1) = 0
[pid 24947] futex(0x7f52702edd54, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 1, {1357409542, 41914000}, ffffffff <unfinished ...>
[pid 24938] <... futex resumed> ) = -1 ETIMEDOUT (Connection timed out)
[pid 24938] futex(0x7f5270102428, FUTEX_WAKE_PRIVATE, 1) = 0
[pid 24938] futex(0x7f5270102454, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 1, {1357409542, 89241000}, ffffffff <unfinished ...>
[pid 24947] <... futex resumed> ) = -1 ETIMEDOUT (Connection timed out)
Last edited by tcn (2013-01-06 17:37:36)
Offline
One more googling: https://bugzilla.redhat.com/show_bug.cgi?id=888380
Last edited by tcn (2013-01-06 17:37:48)
Offline
Same issue here. I experience the same behaviour from kernel version 3.6.8 up to the current one : "3.6.11-1-ARCH #1 SMP PREEMPT Tue Dec 18 08:57:15 CET 2012 x86_64 GNU/Linux".
I tried OpenJDK 1.7 and Oracle JDK with same results. Changing the JVM allocated memory helped a bit by reducing the blocked phase duration. BTW, on Fedora, some users experience the same issue : http://forums.fedoraforum.org/showthrea … 246&page=2
Offline