Hide Forgot
Description of problem: SBD is not a super active upstream project but recently there had been a couple of interesting changes that interfere with a lot of files like stuff for pacemaker-remote and clarification of licensing in most of the files. So it is much cleaner and easier to support to base on the upstream version and patch what we want differently (like e.g. not supporting block-devices for now) instead of doing it the other way round.
In our internal testing the following scenarios do work: * sbd as the only mean of fencing (internally: pacemaker,recovery,recovery-all,sbd,sbd-only,kill_sysrq_panic) * sbd as the secondary fencing, where the primary fencing works (internally: pacemaker,recovery,recovery-all,sbd,sbd-with-other-fencing,kill_sysrq_panic) * sbd as the secondary fencing, where the primary fencing never works (/bin/false) (internally: pacemaker,recovery,recovery-all,sbd,sbd-with-fake-fencing,kill_sysrq_panic) It means that the following expected recovery behaviour of the cluster occurs in the follwing scenarios: * killing node where the resource is active * causing one node to fall off the network * causing switch failure (nodes do not see each other) * killing random (one or more) node: * less than quorum * more than quorum * killing pacemaker on one or more nodes Marking verified.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHBA-2016-2306.html