[sheepdog] [PATCH 0/2] cleanup stale objects

Liu Yuan namei.unix at gmail.com
Mon Aug 27 05:47:53 CEST 2012


On 08/26/2012 03:54 PM, MORITA Kazutaka wrote:
> Next, I'll fix problems in plain store which I found when testing this
> series.
> 
> MORITA Kazutaka (2):
>   tests: add check of stale object cleanup to 032 and 033
>   sheep: remove stale objects after recovery fully completed
> 
>  include/internal_proto.h |    1 +
>  sheep/ops.c              |   54 +++++++++++++++++++++++++++++++++++++++++++--
>  sheep/plain_store.c      |    2 +-
>  sheep/recovery.c         |   30 ++++++++++++++++++++++++-
>  sheep/sheep_priv.h       |    4 +-
>  tests/032                |    1 +
>  tests/032.out            |   15 ++++++++++++
>  tests/033                |    1 +
>  tests/033.out            |   15 ++++++++++++
>  9 files changed, 116 insertions(+), 7 deletions(-)
> 

This patch set might cause problems on 026, the cluster get hangs
sometimes by following bash:

$ while test "$?" -eq 0;do sudo ./check 26 -corosync;done

(gdb) bt
#0  0x00007fccb0458d1d in read () at ../sysdeps/unix/syscall-template.S:82
#1  0x00007fccb04672de in eventfd_read (fd=16, value=0x7fccaf978df8) at
../sysdeps/unix/sysv/linux/eventfd_read.c:27
#2  0x0000000000408ea7 in exec_local_req (rq=0x7fccaf978e30,
data=0x629f58) at request.c:439
#3  0x000000000040ffb7 in notify_recovery_completion_work
(work=0x18f07f0) at recovery.c:352
#4  0x000000000040d155 in worker_routine (arg=<value optimized out>) at
work.c:171
#5  0x00007fccb0d9b971 in start_thread (arg=<value optimized out>) at
pthread_create.c:304
#6  0x00007fccb0466f3d in clone () at
../sysdeps/unix/sysv/linux/x86_64/clone.S:112
#7  0x0000000000000000 in ?? ()

Thanks,
Yuan



More information about the sheepdog mailing list