ceph client mount and backend question
petasanrd911
19 Posts
June 22, 2021, 6:41 pmQuote from petasanrd911 on June 22, 2021, 6:41 pmI have a 5 node cluster and initially I setup the cif, management and backend on the same nic eth0. I added another nic eth1. I would like to move the backend to eth1 nic, is that possible?
also for mounting ceph filesystem on my linux host, which nic does it use? backend ip/nic I presume since there is no configuration page like there is for iscsi, cifs, nfs.
I have a 5 node cluster and initially I setup the cif, management and backend on the same nic eth0. I added another nic eth1. I would like to move the backend to eth1 nic, is that possible?
also for mounting ceph filesystem on my linux host, which nic does it use? backend ip/nic I presume since there is no configuration page like there is for iscsi, cifs, nfs.
admin
2,930 Posts
June 22, 2021, 9:47 pmQuote from admin on June 22, 2021, 9:47 pmYou can change the backend interface from /opt/petasan/config/cluster_info.json on all nodes. You can also move the service interfaces from the ui
clients access services using their ips which should be on the same subnet as the individual services else they would be routed to the gateway defined on the client. the interface being used on the client is the interface having the ip on that subnet.
If you wish to mount/access ceph directly from a a linux client, you need to define a /etc/ceph/ceph.conf + copy access key to /etc/ceph, the conf file lists the ips of the monitors which are on the backend network, your client should have an ip assigned on the backend network to access ceph directly, the interface used will be the one with the client ip on the backend subnet.
You can change the backend interface from /opt/petasan/config/cluster_info.json on all nodes. You can also move the service interfaces from the ui
clients access services using their ips which should be on the same subnet as the individual services else they would be routed to the gateway defined on the client. the interface being used on the client is the interface having the ip on that subnet.
If you wish to mount/access ceph directly from a a linux client, you need to define a /etc/ceph/ceph.conf + copy access key to /etc/ceph, the conf file lists the ips of the monitors which are on the backend network, your client should have an ip assigned on the backend network to access ceph directly, the interface used will be the one with the client ip on the backend subnet.
petasanrd911
19 Posts
June 23, 2021, 3:39 pmQuote from petasanrd911 on June 23, 2021, 3:39 pmthanks for the reply
I presume I make the interface changes on each individual node /opt/petasan/config/cluster_info.json file then restart the cluster? or does changing it on one get replicated to the others?
thanks for the reply
I presume I make the interface changes on each individual node /opt/petasan/config/cluster_info.json file then restart the cluster? or does changing it on one get replicated to the others?
admin
2,930 Posts
June 23, 2021, 6:49 pmQuote from admin on June 23, 2021, 6:49 pmIt does not get replicated and needs to be done on all existing nodes, it is a core network settings cannot be synced as it requites connectivity to sync.
New nodes joining existing cluster will get the new config.
It does not get replicated and needs to be done on all existing nodes, it is a core network settings cannot be synced as it requites connectivity to sync.
New nodes joining existing cluster will get the new config.
ceph client mount and backend question
petasanrd911
19 Posts
Quote from petasanrd911 on June 22, 2021, 6:41 pmI have a 5 node cluster and initially I setup the cif, management and backend on the same nic eth0. I added another nic eth1. I would like to move the backend to eth1 nic, is that possible?
also for mounting ceph filesystem on my linux host, which nic does it use? backend ip/nic I presume since there is no configuration page like there is for iscsi, cifs, nfs.
I have a 5 node cluster and initially I setup the cif, management and backend on the same nic eth0. I added another nic eth1. I would like to move the backend to eth1 nic, is that possible?
also for mounting ceph filesystem on my linux host, which nic does it use? backend ip/nic I presume since there is no configuration page like there is for iscsi, cifs, nfs.
admin
2,930 Posts
Quote from admin on June 22, 2021, 9:47 pmYou can change the backend interface from /opt/petasan/config/cluster_info.json on all nodes. You can also move the service interfaces from the ui
clients access services using their ips which should be on the same subnet as the individual services else they would be routed to the gateway defined on the client. the interface being used on the client is the interface having the ip on that subnet.
If you wish to mount/access ceph directly from a a linux client, you need to define a /etc/ceph/ceph.conf + copy access key to /etc/ceph, the conf file lists the ips of the monitors which are on the backend network, your client should have an ip assigned on the backend network to access ceph directly, the interface used will be the one with the client ip on the backend subnet.
You can change the backend interface from /opt/petasan/config/cluster_info.json on all nodes. You can also move the service interfaces from the ui
clients access services using their ips which should be on the same subnet as the individual services else they would be routed to the gateway defined on the client. the interface being used on the client is the interface having the ip on that subnet.
If you wish to mount/access ceph directly from a a linux client, you need to define a /etc/ceph/ceph.conf + copy access key to /etc/ceph, the conf file lists the ips of the monitors which are on the backend network, your client should have an ip assigned on the backend network to access ceph directly, the interface used will be the one with the client ip on the backend subnet.
petasanrd911
19 Posts
Quote from petasanrd911 on June 23, 2021, 3:39 pmthanks for the reply
I presume I make the interface changes on each individual node /opt/petasan/config/cluster_info.json file then restart the cluster? or does changing it on one get replicated to the others?
thanks for the reply
I presume I make the interface changes on each individual node /opt/petasan/config/cluster_info.json file then restart the cluster? or does changing it on one get replicated to the others?
admin
2,930 Posts
Quote from admin on June 23, 2021, 6:49 pmIt does not get replicated and needs to be done on all existing nodes, it is a core network settings cannot be synced as it requites connectivity to sync.
New nodes joining existing cluster will get the new config.
It does not get replicated and needs to be done on all existing nodes, it is a core network settings cannot be synced as it requites connectivity to sync.
New nodes joining existing cluster will get the new config.