Re: [ClusterLabs] resources cluster stoped with one node
Dne 20. 03. 24 v 23:56 Ken Gaillot napsal(a): On Wed, 2024-03-20 at 23:29 +0100, mierdatutis mi wrote: HI, I've configured a cluster of two nodes. When I start one node only I see that the resources won't start. Hi, In a two-node cluster, it is not safe to start resources until the nodes have seen each other once. Otherwise, there's no way to know whether the other node is unreachable because it is safely down or because communication has been interrupted (meaning it could still be running resources). Corosync's two_node setting automatically takes care of that by also enabling wait_for_all. If you are certain that the other node is down, you can disable wait_for_all in the Corosync configuration, start the node, then re-enable wait_for_all. If you are using pcs, you can achieve the same thing easily by running 'pcs quorum unblock' on the online node. Regards, Tomas [root@nodo1 ~]# pcs status --full Cluster name: mycluster Stack: corosync Current DC: nodo1 (1) (version 1.1.23-1.el7-9acf116022) - partition WITHOUT quorum Last updated: Wed Mar 20 23:28:45 2024 Last change: Wed Mar 20 19:33:09 2024 by root via cibadmin on nodo1 2 nodes configured 3 resource instances configured Online: [ nodo1 (1) ] OFFLINE: [ nodo2 (2) ] Full list of resources: Virtual_IP (ocf::heartbeat:IPaddr2): Stopped Resource Group: HA-LVM My_VG (ocf::heartbeat:LVM-activate): Stopped My_FS (ocf::heartbeat:Filesystem):Stopped Node Attributes: * Node nodo1 (1): Migration Summary: * Node nodo1 (1): Fencing History: PCSD Status: nodo1: Online nodo2: Offline Daemon Status: corosync: active/enabled pacemaker: active/enabled pcsd: active/enabled Do you know what these behaviors are? Thanks ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/ ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
Re: [ClusterLabs] resources cluster stoped with one node
On Wed, 2024-03-20 at 23:29 +0100, mierdatutis mi wrote: > HI, > I've configured a cluster of two nodes. > When I start one node only I see that the resources won't start. Hi, In a two-node cluster, it is not safe to start resources until the nodes have seen each other once. Otherwise, there's no way to know whether the other node is unreachable because it is safely down or because communication has been interrupted (meaning it could still be running resources). Corosync's two_node setting automatically takes care of that by also enabling wait_for_all. If you are certain that the other node is down, you can disable wait_for_all in the Corosync configuration, start the node, then re-enable wait_for_all. > > [root@nodo1 ~]# pcs status --full > Cluster name: mycluster > Stack: corosync > Current DC: nodo1 (1) (version 1.1.23-1.el7-9acf116022) - partition > WITHOUT quorum > Last updated: Wed Mar 20 23:28:45 2024 > Last change: Wed Mar 20 19:33:09 2024 by root via cibadmin on nodo1 > > 2 nodes configured > 3 resource instances configured > > Online: [ nodo1 (1) ] > OFFLINE: [ nodo2 (2) ] > > Full list of resources: > > Virtual_IP (ocf::heartbeat:IPaddr2): Stopped > Resource Group: HA-LVM > My_VG (ocf::heartbeat:LVM-activate): Stopped > My_FS (ocf::heartbeat:Filesystem):Stopped > > Node Attributes: > * Node nodo1 (1): > > Migration Summary: > * Node nodo1 (1): > > Fencing History: > > PCSD Status: > nodo1: Online > nodo2: Offline > > Daemon Status: > corosync: active/enabled > pacemaker: active/enabled > pcsd: active/enabled > > Do you know what these behaviors are? > Thanks > ___ > Manage your subscription: > https://lists.clusterlabs.org/mailman/listinfo/users > > ClusterLabs home: https://www.clusterlabs.org/ -- Ken Gaillot ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
