On 11/06/2013, at 9:28 PM, andreas graeper <agrae...@googlemail.com> wrote:
> hi, > i think this is a question related to recent one discussed in this thread: > i use ocf:heartbeat:symlink > now i use > crm <<EOL > primitive > order > colocation > EOL > to avoid the primitive is started on wrong node before constraints getting > active. > > /mnt/mirror is mountpoint of mirrored blockdevice > /var/mail -> /mnt/mirror/var/mail > > the link is created / removed as expected on master-node ( drbd:master ) > but crm_mon shows : > > Failed actions: > p_lnsmail_monitor_0 (node=n2, call=417, rc=5, status=complete): not > installed > > 1) the link gets created > 2) the target cannot exists on drbd:slave, cause it is on /dev/drbd0 > > whats wrong ? what kind of check fails ? We are checking that "its not running somewhere it shouldn't be before starting it where it should". This happens before _any_ resources are started, including your symlink resource. > thanks > andreas > > > > 2013/6/7 Andrew Beekhof <and...@beekhof.net> > > On 07/06/2013, at 2:52 AM, andreas graeper <agrae...@googlemail.com> wrote: > > > > > thanks awfully for all your answers. > > i started reading about ha/cluster/drbd/pacemaker/.. two weeks ago. it is > > not yet easy to precise my questions. > > > > different os on test-environment only, but i already decided to have equal > > os for testing, too. > > crm configure <<EOF .. # great !!! > > > > my situation: two Nodes n1,n2 and all services (that must not run twice, > > and are managed by pacemaker) run on one node only. the other > > shall take over in case the first fails. cause both machines are running, > > you may say, that some services run on n1, others on n2. but this moment > > i have only some services that depend on drbd:Master + IPaddr + Filesystem, > > so they have to run on one node. the rest (for today) is simply running on > > both nodes > > or not (pacemaker is not aware of). > > > > but a new question (to come closer to understand in full depth ..รถ..) > > when using location-constraints (more than two nodes and on one node a > > service (its binary) does not exists, negative location-constraint ) > > would the resource-manager not complain about missing service (' .. not > > installed') ? > > It will still complain because we would still check that node to make _sure_ > its not running somewhere it shouldn't be before starting it where it should. > > > _______________________________________________ > Pacemaker mailing list: Pacemaker@oss.clusterlabs.org > http://oss.clusterlabs.org/mailman/listinfo/pacemaker > > Project Home: http://www.clusterlabs.org > Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf > Bugs: http://bugs.clusterlabs.org > > _______________________________________________ > Pacemaker mailing list: Pacemaker@oss.clusterlabs.org > http://oss.clusterlabs.org/mailman/listinfo/pacemaker > > Project Home: http://www.clusterlabs.org > Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf > Bugs: http://bugs.clusterlabs.org _______________________________________________ Pacemaker mailing list: Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker Project Home: http://www.clusterlabs.org Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf Bugs: http://bugs.clusterlabs.org