LVM-activate [start | stop | monitor | meta-data | validate-all]
This agent manages LVM activation/deactivation work for a given volume group.
It supports the following modes, controlled by the vg_access_mode parameter:
* lvmlockd * system_id * clvmd * tagging
1. There are two possible configuration combinations: lvmlockd+LVM-activate and clvm+LVM-activate. However, it is not possible to use both at the same time!
2. Put all "lvmlockd"/"clvmd" volume groups into auto management by the agent if using the cluster to manage at least one of them. If you manage some manually, the stop action of the lvmlockd agent may fail and the node may get fenced, because some DLM lockspaces might be in use and cannot be closed automatically.
The standard monitor operation of depth 0 checks if the VG or LV is valid. If you want deeper tests, set OCF_CHECK_LEVEL to 10:
10: read first 1 byte of the underlying device (raw read)
If there are many underlying devs in VG, it will only read one of the devs. This is not perfect solution for detecting underlying devices livable. e.g. iscsi SAN IO timeout will return EIO, and it makes monitor failed.
The volume group name.
(required, string, no default)
If set, only the specified LV will be activated.
(optional, string, no default)
This option decides which solution will be used to protect the volume group in cluster environment. Optional solutions are: lvmlockd, clvmd, system_id and tagging.
(required, string, no default)
The activation mode decides the visibility of logical volumes in the cluster. There are two different modes: "shared" for cluster filesystem and "exclusive" for local filesystem. With "shared", an LV can be activated concurrently from multiple nodes. With "exclusive", an LV can be activated by one node at a time.
This option only has effect on "lvmlockd"/"clvmd" vg_access_mode. For "system_id" and "tagging", they always mean exclusive activation.
(optional, string, default "exclusive")
The tag used for tagging activation mode.
(optional, string, default "pacemaker")
If set, the volume group will be activated partially even with some physical volumes missing. It helps to set to true when using mirrored logical volumes.
(optional, string, default "false")
This resource agent supports the following actions (operations):
Starts the resource. Suggested minimum timeout: 90s.
Stops the resource. Suggested minimum timeout: 90s.
Performs a detailed status check. Suggested minimum timeout: 90s. Suggested interval: 30s.
Retrieves resource agent metadata (internal use only). Suggested minimum timeout: 10s.
Performs a validation of the resource configuration. Suggested minimum timeout: 20s.
Example CRM Shell
The following is an example configuration for a LVM-activate resource using the crm(8) shell:
primitive p_LVM-activate ocf:heartbeat:LVM-activate \ params \ vgname=string \ vg_access_mode=string \ op monitor timeout="90s" interval="30s" depth="0"
The following is an example configuration for a LVM-activate resource using pcs(8)
pcs resource create p_LVM-activate ocf:heartbeat:LVM-activate \ vgname=string \ vg_access_mode=string \ op monitor timeout="90s" interval="30s" depth="0"
ClusterLabs contributors (see the resource agent source for information about individual authors)