Moving conversation to [email protected], which is the right list for HBase questions.
HBase (and many other distributed systems) rely on clocks being synced for various operations to behave as expected. HBase does recommend running ntpd on all nodes to keep time in proper sync. This is why it also does a hard check of > 30s (default config, tweak-able but not seen a really good reason yet). See the configurations for OS that HBase demands if you deploy it, at: http://hbase.apache.org/book.html#os On Thu, Mar 8, 2012 at 12:20 PM, 周文辉 <[email protected]> wrote: > I installed hbase in my hadoop cluster. > when i started hbase and i got an exception about > ClockOutOfSyncException, somthing like this: > > org.apache.hadoop.hbase.ClockOutOfSyncException: > org.apache.hadoop.hbase.ClockOutOfSyncException: > Server slave,60020,1310983722679 has been rejected; Reported time is > too far out of sync with > master. Time difference of 33514ms > max allowed of 30000ms > > I know i need to sync the time. > > But does anybody know why hbase need this sync? > > Thanks very much! -- Harsh J
