You can get the same behaviour by setting longer timeouts and having the RA not return until it decides one way or another that the resource is good or bad. The best way not to have pacemaker perform a premature failover, is to not tell us about failures until you're sure.
On Wed, Aug 29, 2012 at 2:03 AM, Joe Bill <[email protected]> wrote: > I was wondering if it would be possible to introduce a new resource > state, in addition to those already existing in the OCF specifications > (like 0: no error, 1:succes), to indicate a resource is in a "busy" or > in a "(in) transition", or in a "recovery" or "healing" or "autofix" state, > typically when, while in > some kind of maintenance or recovery situation (fsck'ing a filesystem or > rebuilding the indexes of a database) it can not be started, but if an > error staus was returned, an error counter would increment that could > possibly exclude that resource from being used again without manual > intevention. > > When receiving such a "busy" state, pacemaker, when > left without a failover option for that resource, whould regularly probe > thois resource, over and over again, until the probe returns a > "error", where it fails the resource, or a "success" where, if the > resource is stopped, > allows pacemaker to start the resource when it thinks it needs to. > > Thank you all for your comments on the matter. > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
