On 9/21/07, Junko IKEDA <[EMAIL PROTECTED]> wrote: > > but generally speaking, it is not an bug to send a start action to an > > already started resource (this is required to succeed by the LSB and > > OCF spec's) > > > > and in fact it is something that we need to take advantage of for > > groups and clones > > I'm not sure if it's an essential or not but... > noticed the following things; > If the all resources defined as a group can start in unison, > any resources won't start again. > > when I saw a log message like this, Dummy01 started again. > > group_print: Resource Group: grpDummy01 > native_print: Dummy01 (heartbeat::ocf:Dummy): Started prec370d > native_print: Dummy02 (heartbeat::ocf:Dummy): Stopped > native_print: Dummy03 (heartbeat::ocf:Dummy): Stopped > > It seemed that grpDummy01 haven't started yet then. > but grpDummy01_start was prerequisite for Dummy01_start, > so Dummy01_start might be called again after grpDummy01_start. > If grpDummy01_start (start action for the group resource? group body??) can > start before Dummy01, > it might work well. > > > > The attached logs are cib.xml, ha-debug, bz2 file, and results of ptest. > > > I also posted this to bugzilla #1709 > > > > ok, i'll follow up there > Thanks a lot. > It was a duplicated bug to #1648. > I would appreciate if its severity is increased higher.
you do understand though that its not a bug? what the PE is doing is perfectly legal according to the LSB and OCF specs and adds only minimally to the group's startup time while guaranteeing correctness. _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
