On May 13, 2010, at 4:40 PM, Brodie, Kent wrote: > Greetings! I’m hoping someone can enlighten me as to what conceptual thing > I’m doing wrong? > > I have four services, defined as a colocated group. Under normal > circumstances, things seem to work fine-- when I put a node in standby, > everything moves over as it should , and so on. (By the way, this whole > openais/pacemaker setup is really cool, I’m enjoying this thoroughly). > > The one problem I’m having is the situation where one particular service > dies- and keeps on dying. (simulated, for example, by manually killing the > Apache httpd process, etc). When I do this, the local failcount is > incremented (good)-- the service is restarted on the same host (good) – and > life goes on. Until that is, I reach the migration threshold (as below, 5 > attempts). What I’m looking for is, “ok, after 5 failed attempts of this > service, fail over to the other node. This works-- but only partially? > That is, (using the apache example)—the apache service DOES fail over. > That much works. But—nothing else goes along with it! I thought the > “collocation” parameter is supposed to do this-- am I missing something else? > > > > node bennett \ > attributes standby="off" > node caffey \ > attributes standby="on" > primitive Apache ocf:heartbeat:apache \ > params configfile="/usr/local/apache2/conf/httpd.conf" > port="8000" httpd="/usr/local/apache2/bin/httpd" > statusurl="http://127.0.0.1:8000/server-status"; \ > op monitor interval="30s" \ > meta target-role="Started" \ > meta migration-threshold="5" > primitive ClusterIP ocf:heartbeat:IPaddr2 \ > params ip="141.136.224.243" cidr_netmask="24" \ > op monitor interval="30s" \ > meta target-role="Started" > primitive asys ocf:heartbeat:asys \ > op monitor interval="30s" \ > meta target-role="Started" \ > meta migration-threshold="5" > primitive glassfish ocf:heartbeat:glassfish \ > op monitor interval="30s" \ > op start interval="0" timeout="60s" \ > meta target-role="Started" \ > meta migration-threshold="5" > colocation all_services inf: ClusterIP Apache glassfish asys > order asys-after-glassfish inf: glassfish asys > property $id="cib-bootstrap-options" \ > dc-version="1.0.8-9881a7350d6182bae9e8e557cf20a3cc5dac3ee7" \ > cluster-infrastructure="openais" \ > expected-quorum-votes="2" \ > stonith-enabled="false" \ > no-quorum-policy="ignore" \ > last-lrm-refresh="1273782512" > rsc_defaults $id="rsc-options" \ > resource-stickiness="100" >
Hi, You might find more help in pacemaker maillist: [email protected]. But config doesn't look unreasonable. Have you checked how constraints actually look like? cibadmin -Q -o constraints and as a workaround you can define a group instead of using collocation/order group all_services ClusterIP Apache glassfish asys Vadym
_______________________________________________ Openais mailing list [email protected] https://lists.linux-foundation.org/mailman/listinfo/openais
