Why do we need to track such dead servers which are not going to come back. Pardon, maybe I missed some pointer.
Any thought? Regards, Pankaj -----Original Message----- From: mallik.v.ar...@gmail.com [mailto:mallik.v.ar...@gmail.com] Sent: Friday, September 18, 2020 11:19 PM To: dev@hbase.apache.org Subject: Re: [Discuss] Remove server from RSGroup Moving any dead region servers to default rsgroup would remove them from knows server list. On Fri, Sep 18, 2020, 11:08 PM Pankaj Kumar <pankajku...@apache.org> wrote: > Dear Folks, > > Currently we don’t allow user to remove a dead server from a RSGroup. > > > https://github.com/apache/hbase/blob/e386ec566f89162e4aa08229f40a708ea > 3c9a038/hbase-server/src/main/java/org/apache/hadoop/hbase/rsgroup/RSG > roupInfoManagerImpl.java#L899 > > > There can be a scenario where a server is crashed due to hardware > failure and will not come back, admin would be interested in removing > that server from rsgroup which is not allowed. > > > IMO, it should be given a choice to user instead of restricting it. > > > Regards, > > Pankaj >
