I assume you're aware of the trash feature of hdfs See: https://issues.apache.org/jira/browse/HDFS-2740
FYI > On Oct 16, 2015, at 10:31 AM, "[email protected]" > <[email protected]> wrote: > > not a sucure hbase > we delete the tableInfo by mistabke is a case > the other is: if we can not drop a table with drop in hbase shell, so we can > drop table by ourselves, delete hdfs, delete meta, and so on, > so, after we delete hdfs, we need to restart master to refresh the cache. > > so, i think we can add a task to synchronize master' cache and hdfs > > > > [email protected] > > From: Ted Yu > Date: 2015-10-17 01:17 > To: [email protected] > Subject: Re: add a task in master to check if tableInfo is missinig > Are you running hbase on secure cluster ? > > As J-M said, hbck can come to the rescue if you encounter such a case. > > Cheers > > On Fri, Oct 16, 2015 at 10:11 AM, [email protected] < > [email protected]> wrote: > >> >> if we delete the tableInfo file by mistake, then if hbase is restated, the >> regions of this table can not be opend successfully, >> so, can we add a task in master to check its cache and hdfs to check if a >> tableInfo is missing, if missing, we can write a new one using master's >> cache >> >> >> [email protected] >>
