For large numbers of RHEL guests, the performance team recommends taking sched_migration_cost to 10x the base value.
For RHEL, also see: https://bugzilla.redhat.com/show_bug.cgi?id=825222
For RHEV-H images, this is done as part of bug 965304. It's worth noting that this is *only* done for RHEV-H. For RHEL hosts, either vdsm or some other component would need to configure this if needed. If we want to ensure it's done on RHEL hosts automatically, then this should move to a different component. If not, then this can be closed as a duplicate of the above bug.
afaik, we handle this for RHEL hosts as part of host bootstrap, so we do need this for RHEV-H. worth mentioning to QE how to test the actual running value for this
OTOPI is needed only until Bug 825222 will handle it in RHEL.
(In reply to Doron Fediuck from comment #5) > OTOPI is needed only until Bug 825222 will handle it in RHEL. Which doesn't look likely So we need to handle this. Doron, do you know if the virt-host profile handles this already?
(In reply to Andrew Cathrow from comment #6) > (In reply to Doron Fediuck from comment #5) > > OTOPI is needed only until Bug 825222 will handle it in RHEL. > > Which doesn't look likely > So we need to handle this. > > Doron, do you know if the virt-host profile handles this already? Andy, not yet. We should monitor bug 969491 for that.
Closing as dupe as this will be handled in 6.5 BZ 969491 *** This bug has been marked as a duplicate of bug 969491 ***