What services/daemons are available for Red Hat Enterprise Linux Cluster?

Solution Verified - Updated

Environment

  • Red Hat Cluster Suite 4+
  • Red Hat GFS 4+
  • Red Hat Enterprise Linux 5 Advanced Platform (Clustering)
  • Red Hat Enterprise Linux 5 Advanced Platform (Clustering and GFS/GFS2)
  • Red Hat Enterprise Linux Server 6 (with the High Availability Add on)
  • Red Hat Enterprise Linux Server 6 (with the High Availability and Resilient Storage Add Ons)

Issue

  • What services are available in my Red Hat Enterprise Linux Cluster?
  • How do I start and stop Red Hat Enterprise Linux Cluster services?

Resolution

There are several different services available in the Red Hat Clustering that need to be started in order to provide certain functionality. Please note that each version of Red Hat Enterprise Linux(RHEL) will have their own set of services. Some versions of RHEL contain similar services, some contain newer services that deprecate other services.The list of services(daemons) from all versions of RHEL are listed below. All the daemons are located under the directory */etc/init.d/. The information provided below does not contain every service that is provided but lists the most common services:

  • ccsd: is the cluster configuration daemon responsible for parsing /etc/cluster/cluster.conf and managing configuration versions.  In Red Hat Enterprise Linux 5, this service no longer needs to be started independently, but rather is started automatically by the cman service.  ccsd is provided by the ccsd package in RHEL 4, and by the cman package in RHEL 5.

  • cman: is the cluster manager which is responsible for quorum, voting, heartbeats, and communication.  In Red Hat Enterprise Linux 4, cman has both a userspace and kernel component, meaning nodes must have a specific version of the cman-kernel package installed depending on the kernel version that is running.  Both components are available from the Clustering channel in RHEL 4.  In RHEL 5 and RHEL 6 cman resides completely in userspace so only the cman package is needed, which is available from the base channel. This means that RHEL 5 does not have any cman kernel packages.

  • fenced: is the fencing daemon.  It only needs to be started independently on Red Hat Enterprise Linux 4, whereas in RHEL 5 and RHEL 6 it is started by the cman service. In RHEL 4, fenced is provided by the fence package available in the Clustering channel.  In RHEL 5 and RHEL 6, it is provided by the cman package.

  • qdisk: is responsible for managing a quorum disk(qdisk).  A qdisk can be used to bolster quorum by defining additional criteria for a node's membership in the cluster, and also serves as another method of communication for nodes to determine which node should be fenced in the event of a network failure.  More information on qdisk can be found in the This content is not included.Cluster Administration Guide.   The qdiskd daemon is provided by the cman package on RHEL 4, RHEL 5. On RHEL 6 there is no daemon because cman handles starting qdisk.

  • clvmd: is the cluster logical volume manager (LVM) daemon.  Anytime cluster nodes are utilizing shared storage managed by LVM, whether it be ext3, GFS, GFS2, or any other filesytem residing on it, there must be a mechanism for protecting the metadata from concurrent access or corruption.  For example if one node were to read the metadata for a volume group and cache it, then another node were to modify that metadata to reduce the size of a logical volume, the original node would not know about that change and could overwrite critical data outside of the logical volume.  The clvmd daemon manages metadata for any volume group with the clustered flag set, and is provided by the lvm2-cluster package available in the Cluster Storage channel on RHN.  An alternative method for managing LVM on shared storage in a cluster is described in What is a Highly Available LVM (HA-LVM) configuration and how do I implement it?. The daemon clvmd is provided  by the package lvm2-cluster on RHEL 4, RHEL 5, and RHEL 6.

  • gfs and gfs2: services exist to manage mounting and unmounting GFS and GFS2 filesystems from /etc/fstab, respectively. Typically filesystems that are listed in /etc/fstab are mounted by the rc.sysinit program during the boot process or by the netfs service for those that have the _netdev mount option specified.  Neither of these are suitable for clustered filesystems because they start before the cluster infrastructure on which these filesystems depend.  Any gfs or gfs2 filesystem listed in fstab will be mounted when these services are started, and will be unmounted when they are stopped.  The gfs service is provided by the gfs-utils package and gfs2 is provided by gfs2-utils, both in the Cluster Storage channel on RHN.  The daemon gfs is provided  by the package gfs-utils on RHEL 4 and RHEL 5. The daemon gfs2 is provided  by the package gfs2-utils on RHEL 5 and RHEL 6.

  • rgmanager: is the cluster resource group manager, and is responsible for the starting, stopping, relocating, and monitoring of highly-available services managed by the cluster.  Upon starting it, any services marked with autostart="1" in /etc/cluster/cluster.conf will be initialized and started.  Services can then be managed with Conga, system-config-cluster, clustat, and clusvcadm.  This service is provided by the rgmanager package on RHEL4, RHEL5, and RHEL6.

  • cmirrord: cmirrord is the daemon that tracks LVM mirror log information in a cluster. In order to configure LVM mirrors in a cluster, this needs to be running. This is also required for doing pvmove operation in a cluster.

  • luci: is the server component of the Conga administration utility.  It only needs to run on the system that will function as the web fronted for Conga, and is provided by the luci package on RHEL 5 and RHEL 6.

  • ricci: is the client component of the Conga administration utility, and needs to be started on every node in any cluster that will be managed by Conga. It is provided by the ricci package on RHEL 4, RHEL 5 and RHEL 6.

  • scsi_reserve is the start-up script needed for setting up SCSI Persistent Reservations for use with fence_scsi. It is only available in RHEL 4 and 5. RHEL 6 SCSI-PR fencing uses a different mechanism ("unfencing") for setting up reservations/registrations.

Please note that when installing a cluster some packages will install other daemons that are not listed here. These daemons should be turned off at boot time because other services will handle starting them. For example, the services openais on RHEL 5 and RHEL 6 should be disabled. Another example is the service corosync which should be disabled on RHEL 6.

  • If you are looking for the order in which to stop/start these services please refer to this.
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.