On Thu, Jun 28, 2012 at 11:36 AM, Liu Yuan <[email protected]> wrote: > On 06/27/2012 10:05 PM, Liu Yuan wrote: >> Then we can benefit its retry mechanism. > > Applied this series after changing English typos, thanks.
I found a problem caused by this series. As this series use exec_local_req() for pull/push/forward_read_xx/forward_write_xx methods, when the cluster become busy, all io/gateway threads may call exec_local_req() and wait in eventfd_read() at the same time, then there are no other idle threads to do local request, as a result the cluster are blocked by itself. > > Yuan > > -- > sheepdog mailing list > [email protected] > http://lists.wpkg.org/mailman/listinfo/sheepdog -- Yunkai Zhang Work at Taobao -- sheepdog mailing list [email protected] http://lists.wpkg.org/mailman/listinfo/sheepdog
