[sheepdog] [sheepdog-users] [SPAM] [ANNOUNCE, CAUTION] sheepdog stable release v0.8.3_rc0

Hitoshi Mitake mitake.hitoshi at lab.ntt.co.jp
Wed Aug 20 10:39:17 CEST 2014


At Fri, 15 Aug 2014 17:42:31 +0800,
Ruoyu wrote:
> 
> 
> On 2014年08月15日 16:55, Hitoshi Mitake wrote:
> > At Fri, 15 Aug 2014 10:07:33 +0900,
> > Hitoshi Mitake wrote:
> >> At Thu, 14 Aug 2014 17:41:18 +0800,
> >> Ruoyu wrote:
> >>> Functional test 029 is failed in stable branch.
> >>> Maybe one or more patches have not been backported.
> >>>
> >>> Aug 14 17:26:01 EMERG [io 16169] get_store_path(46) PANIC: invalid 
> >>> ec_index 32
> >>> Aug 14 17:26:01 EMERG [io 16169] crash_handler(267) sheep exits 
> >>> unexpectedly (Aborted).
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(833) sheep.c:269: 
> >>> crash_handler
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(847) 
> >>> /lib/x86_64-linux-gnu/libpthread.so.0(+0xfbaf) [0x7fec6c2b7baf]
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(847) 
> >>> /lib/x86_64-linux-gnu/libc.so.6(gsignal+0x36) [0x7fec6b9f8f76]
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(847) 
> >>> /lib/x86_64-linux-gnu/libc.so.6(abort+0x147) [0x7fec6b9fc5e7]
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(833) plain_store.c:46: 
> >>> get_store_path
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(833) plain_store.c:303: 
> >>> default_read
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(833) ops.c:951: peer_read_obj
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(833) ops.c:1431: 
> >>> do_process_work
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(833) work.c:348: 
> >>> worker_routine
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(847) 
> >>> /lib/x86_64-linux-gnu/libpthread.so.0(+0x7f6d) [0x7fec6c2aff6d]
> >>> Aug 14 17:26:01 EMERG [io 16169] sd_backtrace(847) 
> >>> /lib/x86_64-linux-gnu/libc.so.6(clone+0x6c) [0x7fec6babc9cc]
> >> Thanks a lot for your report. I'll pick patches for this problem.
> > The commit 515cc2cc216620cedfe45ea6f5978ddb34d70a90 solves the above
> > problem. But the commit introduces incompatibility because it adds new
> > operation. Unfortunately, the problem (dog vdi object location cannot
> > be used for ec vdi) cannot be solved in stable-0.8...
> You mean the user cannot execute dog vdi object command if the vdi is
> erasure coded?

Yes, seems that the commit 79706e07a06880e4eee32f467685028af85f54f9
introduced the change. And the fix for the problem (including protocol
change which cannot be backported to stable-0.8) is provided after
release of v0.8.0.

Thanks,
Hitoshi

> >
> > Thanks,
> > Hitoshi
> >
> >> Thanks,
> >> Hitoshi
> >>
> >>> On 2014年08月13日 14:54, Hitoshi Mitake wrote:
> >>>> Hi sheepdog users and developers,
> >>>>
> >>>> I released v0.8.3_rc0 of stable branch. You can download a source
> >>>> archive from these URLs:
> >>>> tar.gz: https://github.com/sheepdog/sheepdog/archive/v0.8.3_rc0.tar.gz
> >>>> zip: https://github.com/sheepdog/sheepdog/archive/v0.8.3_rc0.zip
> >>>>
> >>>> This update contains important bugfixes, especially related to
> >>>> zookeeper driver. In addition, enhancements of log messages are
> >>>> backported.
> >>>>
> >>>> I have to emphasis on the point that updating from v0.8.x (x < 3) to
> >>>> v0.8.3_rc0 requires "-u" option, because it contains updating store
> >>>> format. Sorry for inconvenience but the format update is a necessary
> >>>> change for clusters which store many objects.
> >>>>
> >>>> If no one disagrees about this release candidate in 2 days, it will be
> >>>> v0.8.3 officially.
> >>>>
> >>>> Below is a summary of commits:
> >>>>
> >>>> Meng An (1):
> >>>>        sheep: remove last slash for new->path in md_add_disk()
> >>>>
> >>>> Ruoyu (8):
> >>>>        sheep/zookeeper: ensure zookeeper is connected
> >>>>        zk_control: add list and purge queue nodes
> >>>>        zookeeper: fix zk_event structure not initialized
> >>>>        sheep/ops: fix incorrect error log on remove_epoch
> >>>>        sheep/request: straightforward error message in sheep_exec_req
> >>>>        lib/event: fix and unify error message
> >>>>        sheep/cluster: remove ephemeral node when zookeeper master leaves
> >>>>        sheep/request: fix fd leak
> >>>>
> >>>> Thanks,
> >>>> Hitoshi
> >>>
> >>> -- 
> >>> sheepdog mailing list
> >>> sheepdog at lists.wpkg.org
> >>> http://lists.wpkg.org/mailman/listinfo/sheepdog
> 
> 
> -- 
> sheepdog-users mailing lists
> sheepdog-users at lists.wpkg.org
> http://lists.wpkg.org/mailman/listinfo/sheepdog-users



More information about the sheepdog mailing list