Strahil
2019-Sep-30 06:06 UTC
[Gluster-users] Where does Gluster capture the hostnames from?
In replicated volumes you can use either reset-brick or replace-brick. Still you will have to heal all the data ... Which for large volumes will take a lot of time. Best Regards, Strahil NikolovOn Sep 30, 2019 06:51, TomK <tomkcpr at mdevsys.com> wrote:> > Because this was a LAB I could quickly remove the gluster setup and > recreated it and used the FQDN's, it quickly picked up the new names. > Exactly as expected per this thread. > > [root at mdskvm-p02 network-scripts]# gluster volume status > Status of volume: mdsgv01 > Gluster process???????????????????????????? TCP Port? RDMA Port? Online? Pid > ------------------------------------------------------------------------------ > Brick mdskvm-p01.nix.mds.xyz:/mnt/p01-d01/g > lusterv01?????????????????????????????????? 49152???? 0????????? Y > 4375 > Brick mdskvm-p02.nix.mds.xyz:/mnt/p02-d01/g > lusterv02?????????????????????????????????? 49152???? 0????????? Y > 4376 > NFS Server on localhost???????????????????? N/A?????? N/A??????? N?????? N/A > Self-heal Daemon on localhost?????????????? N/A?????? N/A??????? Y > 4402 > NFS Server on mdskvm-p01.nix.mds.xyz??????? N/A?????? N/A??????? N?????? N/A > Self-heal Daemon on mdskvm-p01.nix.mds.xyz? N/A?????? N/A??????? Y > 4384 > > Task Status of Volume mdsgv01 > ------------------------------------------------------------------------------ > There are no active volume tasks > > [root at mdskvm-p02 network-scripts]# > > Would be handy to have a rename function in future releases. > > Cheers, > TK > > On 9/25/2019 7:47 AM, TomK wrote: > > Thanks Thorgeir.? Since then I upgraded to Gluster 6.? Though this issue > > remaind the same, anything in the way of new options to change what's > > displayed? > > > > Reason for the ask is that this get's inherited by oVirt when doing > > discovery of existing gluster volumes.? So now I have an IP for a host, > > a short name for a host and FQDN's for the rest. > > > > > > [root at mdskvm-p02 glusterfs]# gluster volume status > > Status of volume: mdsgv01 > > Gluster process???????????????????????????? TCP Port? RDMA Port? Online? > > Pid > > ------------------------------------------------------------------------------ > > > > Brick mdskvm-p02.nix.mds.xyz:/mnt/p02-d01/g > > lusterv02?????????????????????????????????? 49152???? 0????????? Y 22368 > > Brick mdskvm-p01.nix.mds.xyz:/mnt/p01-d01/g > > lusterv01?????????????????????????????????? 49152???? 0????????? Y 24487 > > NFS Server on localhost???????????????????? N/A?????? N/A??????? N?????? > > N/A > > Self-heal Daemon on localhost?????????????? N/A?????? N/A??????? Y 22406 > > NFS Server on 192.168.0.60????????????????? N/A?????? N/A??????? N?????? > > N/A > > Self-heal Daemon on 192.168.0.60??????????? N/A?????? N/A??????? Y 25867 > > > > Task Status of Volume mdsgv01 > > ------------------------------------------------------------------------------ > > > > There are no active volume tasks > > > > [root at mdskvm-p02 glusterfs]# > > > > Cheers, > > TK > > > > On 9/24/2019 2:58 AM, Thorgeir Marthinussen wrote: > >> In an effort to answer the actual question, in my experience the > >> Gluster internals captures the address the first time you probe > >> another node. > >> So if you're logged into the first node and probe the second using an > >> IP-address, that is what will "forever" be displayed by gluster > >> status, and if you use a hostname that's what will be shown. > >> Brick paths are captured when the brick is registered, so using a path > >> with IP will always show the IP as part of the path, and hostname will > >> show that, etc. > >> > >> I haven't verified, but the second node I believe will attempt a > >> reverse lookup of the first node (when probing first->second) and > >> record that name (if any) as the "primary" name of the first node. > >> Also good to know, nodes can have multiple names, the primary name is > >> the one "