Ce contenu n'est pas disponible dans la langue sélectionnée.
Chapter 39. Clustering
The pcs tool now manages bundle resources in Pacemaker
As a Technology Preview starting with Red Hat Enterprise Linux 7.4, Pacemaker supports a special syntax for launching a Docker container with any infrastructure it requires: the bundle. After you have created a Pacemaker bundle, you can create a Pacemaker resource that the bundle encapsulates. For information on Pacemaker support for containers, see the High Availability Add-On Reference guide.
There is one exception to this feature being Technology Preview: As of RHEL 7.4, Red Hat fully supports the usage of Pacemaker bundles for Red Hat Openstack Platform (RHOSP) deployments. (BZ#1433016)
New fence-agents-heuristics-ping
fence agent
As a Technology Preview, Pacemaker now supports the
fence_heuristics_ping
agent. This agent aims to open a class of experimental fence agents that do no actual fencing by themselves but instead exploit the behavior of fencing levels in a new way.
If the heuristics agent is configured on the same fencing level as the fence agent that does the actual fencing but is configured before that agent in sequence, fencing issues an
off
action on the heuristics agent before it attempts to do so on the agent that does the fencing. If the heuristics agent gives a negative result for the off
action it is already clear that the fencing level is not going to succeed, causing Pacemaker fencing to skip the step of issuing the off
action on the agent that does the fencing. A heuristics agent can exploit this behavior to prevent the agent that does the actual fencing from fencing a node under certain conditions.
A user might want to use this agent, especially in a two-node cluster, when it would not make sense for a node to fence the peer if it can know beforehand that it would not be able to take over the services properly. For example, it might not make sense for a node to take over services if it has problems reaching the networking uplink, making the services unreachable to clients, a situation which a ping to a router might detect in that case. (BZ#1476401)
Heuristics supported in corosync-qdevice
as a Technology Preview
Heuristics are a set of commands executed locally on startup, cluster membership change, successful connect to
corosync-qnetd
, and, optionally, on a periodic basis. When all commands finish successfully on time (their return error code is zero), heuristics have passed; otherwise, they have failed. The heuristics result is sent to corosync-qnetd
where it is used in calculations to determine which partition should be quorate. (BZ#1413573, BZ#1389209)
New LVM and LVM lock manager resource agents
As a Technology Preview, Red Hat Enterprise Linux 7.6 introduces two new resource agents:
lvmlockd
and LVM-activate
.
The
LVM-activate
agent provides a choice from multiple methods for LVM management throughout a cluster:
- tagging: the same as tagging with the existing
lvm
resource agent - clvmd: the same as clvmd with the existing
lvm
resource agent - system ID: a new option for using system ID for volume group failover (an alternative to tagging).
- lvmlockd: a new option for using
lvmlockd
anddlm
for volume group sharing (an alternative toclvmd
).
The new
lvmlockd
resource agent is used to start the lvmlockd
daemon when LVM-activate
is configured to use lvmlockd
.
For information on the lvmlockd and LVM-activate resource agent, see the PCS help screens for those agents. For information on setting up LVM for use with
lvmlockd
, see the lvmlockd(8)
man page. (BZ#1513957, BZ#1634729)