Another question while I continue testing. How does one start the cluster with only one node? I'm having an issue with one of my cluster set-ups trying to start it. It is a drbd/nfs/mysql type with virtual ip. When I try to start the cluster on a single machine, none of the services will start. I expect to be able to start the cluster software, and be able to force it to be a single part of the cluster. I can do this on the command line (bring up drbd, force it to be primary, mount, bring up file system and mysql, and lastly virtual ip). Then later on when the second node comes up drbd will attach, sync up, and all is well (but no cluster manager obviously). So, inside crm, how does one make drbd become primary and stay that way? If I bring my cluster config up, drbd says WFConnection: secondary/unknown. If I force drbd to primary at normal command line, drbd will briefly switch, but then CRM puts it back to secondary and things don't work again. Is there a way to bring up the cluster on only a single node, for emergency or maintenance reasons, without the other node being up?
kevins7189 wrote: > > > > Ulrich Windl wrote: >> >>>>> kevins7189 <[email protected]> schrieb am 15.09.2011 um 15:22 in >>>>> Nachricht >> <[email protected]>: >> >>>If "disturbing" means "re balancing of resources", "stickiness" is the word you may be looking for. >> I have resource stickiness set to 100 as listed in most guides. Is that >> too low? No matter what stickiness is set to, I don't understand how >> that applies to a node becoming available and its need to do anything but >> participate and not mess with the already running resources. >> >> >>>Distinguish between "node locks up", "network locks up" and "resource monitor times out" >> My first encounter was with a node losing access to the file system due >> to load and turned out to be a BIOS issue, network communication was >> still up, but the nodes decided nothing could run, even though one node >> was perfectly fine and already was the master. I guess what I would like >> to happen is for the cluster and services to run on a working node "no >> matter what", as in I can deal with the possibility of corrupt data >> better than a total cluster outage rendering nfs and mysql unavailable. >> This didn't seem to be an issue with heartbeat 2.1, I could force it to >> do what I wanted to if needed. >> >> I think I have a breakthrough today on STONITH over DRAC, but need to >> reconfigure everything for support, annoying that DELL makes you bind the >> onboard nics just to make it work. >> >>> >>> Here is my configuration, is there anything here that is not correct for >>> a >>> two node cluster with drbd/nfs/mysql? I followed the >>> drbd/nfs/mysql/postgesql guide, but eliminated postgesql and the need >>> for >>> two drbd partitions because they are not needed. My expectation is that >>> if >>> one node goes down, the other node takes over, that's it. >>> >>> Here is my config if someone would be so kind as to check it out. >>> >>> node node0-1 \ >>> attributes standby="off" >>> node node0-2 \ >>> attributes standby="off" >>> primitive drbd_nfs ocf:linbit:drbd \ >>> params drbd_resource="node0-drbd" \ >>> op monitor interval="15s" >>> primitive fs_nfs ocf:heartbeat:Filesystem \ >>> params device="/dev/drbd0" directory="/pub" fstype="ext3" \ >>> op start interval="0" timeout="60" \ >>> op stop interval="0" timeout="120" \ >>> meta target-role="Started" >>> primitive ip1 ocf:heartbeat:IPaddr2 \ >>> params ip="192.168.1.2" nic="eth0" \ >>> op monitor interval="5s" >>> primitive ip1arp ocf:heartbeat:SendArp \ >>> params ip="192.168.1.2" nic="eth0" >>> primitive mysql ocf:heartbeat:mysql \ >>> params binary="/usr/local/mysql/bin/mysqld_safe" >>> config="/etc/my.cnf" datadir="/usr/local/mysql/mysql-data" >>> socket="/var/run/mysql/mysql.sock" pid="/var/run/mysql/mysqld.pid" \ >>> op start interval="0" timeout="120" \ >>> op stop interval="0" timeout="120" \ >>> op monitor interval="60" timeout="30" depth="0" \ >>> meta target-role="Started" >>> primitive nfs lsb:nfsserver \ >>> meta target-role="Started" >>> group HAServices ip1 ip1arp fs_nfs nfs mysql \ >>> meta target-role="Started" >>> ms ms_drbd_nfs drbd_nfs \ >>> meta master-max="1" master-node-max="1" clone-max="2" >>> clone-node-max="1" notify="true" >>> colocation ms-drbd-nfs-with-haservices inf: ms_drbd_nfs:Master >>> HAServices >>> order fs-nfs-before-nfs inf: fs_nfs:start nfs:start >>> order ip-before-arp inf: ip1:start ip1arp:start >>> order ip-before-ms-drbd-nfs inf: ip1:start ms_drbd_nfs:promote >>> order ms-drbd-nfs-before-fs-nfs inf: ms_drbd_nfs:promote fs_nfs:start >>> order ms-drbd-nfs-before-mysql inf: ms_drbd_nfs:promote mysql:start >>> property $id="cib-bootstrap-options" \ >>> dc-version="--" \ >>> cluster-infrastructure="openais" \ >>> expected-quorum-votes="2" \ >>> stonith-enabled="false" \ >>> no-quorum-policy="ignore" \ >>> cluster-recheck-interval="0" >>> rsc_defaults $id="rsc-options" \ >>> resource-stickiness="100" >>> >>> Also, I could have swore I read a post where someone was able to get >>> STONITH >>> to work by ssh'ing to the other node and issue an ipmi shutdown that >>> way, I >>> think that would solve my issue if I really need STONITH. >>> >>> thanks in advance for any assistance. >> >> >> >> >> _______________________________________________ >> Linux-HA mailing list >> [email protected] >> http://lists.linux-ha.org/mailman/listinfo/linux-ha >> See also: http://linux-ha.org/ReportingProblems >> >> > > -- View this message in context: http://old.nabble.com/First-configuration-Issues-tp32467174p32494822.html Sent from the Linux-HA mailing list archive at Nabble.com. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
