Which configuration changes require a restart in a Red Hat Enterprise Linux cluster?

Solution Verified - Updated

Environment

  • Red Hat Enterprise Linux Advanced Platform 5+ (Clustering and GFS/GFS2)
  • Red Hat Enterprise Linux 5 with High Availability or Resilient Storage Add-on
  • Red Hat Enterprise Linux 6 with High Availability or Resilient Storage Add-on

Issue

  • What cluster configuration changes require a restart of all RHEL cluster services?
  • Do I need to restart my entire cluster after making configuration changes?
  • I have modified and propagated my cluster.conf in two node cluster, do I need to restart entire cluster for changes to take effect or this can be done online?

Resolution

After changing the cluster configuration in /etc/cluster/cluster.conf and then propagating the changes to all the cluster nodes, the following changes require fully stopping then restarting the cluster software:

  • Adding or removing the cman two_node variable
  • Renaming the cluster
  • Changing "upgrading" mode in RHEL6 [unsupported]
  • Changing any openais timers in RHEL 5 (see footnote for RHEL 6) [1]
  • Changing qdiskd device [2]
  • Changing any qdiskd timers [2]
  • Adding, changing or deleting heuristics for qdiskd [2]
  • Changing central_processing mode for rgmanager [3]
  • Changing the cluster multicast address or port
  • Altering cluster transport mode (multicast, broadcast, or UDP unicast)
  • Changing post_join_delay or post_fail_delay (for RHEL5 cman prior to cman-2.0.115-34.el5 only)
  • All cman parameters except for expected_votes, max_queued, and items in the <clusternodes .../> section of /etc/cluster/cluster.conf [4]
  • Introducing new nodes into a cluster utilizing the udpu transport

[1] a timer is any value specified in the <totem/> element of /etc/cluster/cluster.conf when using RHEL 5. When using RHEL6 or RHEL7, the totem timeout is active after propagation without a service restart.
[2] In RHEL 4 and RHEL 5 Update 4 and earlier, the entire cluster must be restarted for these changes to take effect. In RHEL 5 Update 5 and later, a global stop of qdiskd, update config, global start of qdisk is required. To stop/start qdiskd in RHEL 5 and 6, refer How do I restart the quorum disk daemon (qdiskd)?
[3] global stop of rgmanager, update config, global start of rgmanager is required
[4] for instance, changing quorum_dev_poll requires a full cluster restart

SBR
Product(s)
Components
Category
Tags

This solution is part of Red Hat’s fast-track publication program, providing a huge library of solutions that Red Hat engineers have created while supporting our customers. To give you the knowledge you need the instant it becomes available, these articles may be presented in a raw and unedited form.