ClusterLabs/pacemaker Pacemaker-2.1.10-rc1

Pacemaker 2.1.10 - Release Candidate 1

on GitHub

Fixes since Pacemaker-2.1.9

• build: Fix default pacemaker-remoted path on Fedora >= 42
• daemons: Fix a bug iterating in get_op_total_timeout.
• daemons: Fix a toctou bug when fetching additional schemas.
• executor: avoid use-after-free upon shutdown
• libcrmcluster: better detect remote nodes in peer cache
• libcrmcluster: use pcmk__cluster_get_xml_id() when possible
• libcrmservice: Fix overrides for systemd resources
• libcrmservice: Don't leak msg if systemd_proxy is NULL.
• libpacemaker: set fail-count to INFINITY for fatal failures
• libpe_status: consider parents of an unmanaged resource active on the node
• pacemaker-attrd: bail earlier if value won't be written
• pacemaker-attrd: make a peer learn our node name once it has joined
• pacemaker-attrd: prevent segfault if a peer leaves when its name is unknown yet
• pacemaker-attrd: remember names of peers from attribute update in case unknown
• pacemaker-attrd: track node XML IDs independent of attribute values
• pacemaker-attrd: use API to get peer XML ID
• pacemaker-remoted: improve exit codes for schema failures
• scheduler: Don't always fence online remote nodes, and require a cluster
  option to enable this new fencing behavior
• various: clean up library memory at child exit
• various: Correctly detect completion of systemd start/stop actions

Public API changes since Pacemaker-2.1.9

• libcrmcommon: add pcmk_common_cleanup()