SUNW.qfs - Resource type implementation for the Oracle HSM shared file system in a Solaris Cluster environment
SUNWqfs
The SUNW.qfs
resource type implementation supports the
Oracle HSM shared file system installed in a Solaris Cluster
environment.
It defines a failover resource for the shared file system's
metadata server (MDS).
did
devices in the mcf
(4) file.
Additionally, the Solaris Cluster node-private hostnames (see scconf
(1M))
must be used in the shared hosts file (see hosts.fs
(4))
for the host IP addresses.
See r_properties
(5) for a complete description of the
following resource properties:
Validate_timeout Minimum: 60 Default: 180 Boot_timeout Minimum: 30 Default: 300 Prenet_start_timeout Minimum: 60 Default: 360 Start_timeout Minimum: 60 Default: 120 Stop_timeout Minimum: 60 Default: 120 Postnet_stop_timeout Minimum: 60 Default: 120 Monitor_start_timeout Minimum: 60 Default: 120 Monitor_stop_timeout Minimum: 60 Default: 120 Monitor_check_timeout Minimum: 60 Default: 120 Init_timeout Minimum: 30 Default: 300 Fini_timeout Minimum: 60 Default: 120 Update_timeout Minimum: 60 Default: 120 Retry_Count Maximum: 10 Default: 2 Tunable: Anytime Retry_Interval Maximum: 3600 Default: 750 Tunable: Anytime Thorough_probe_interval Maximum: 3600 Default: 60 Tunable: Anytime FailOver_Mode Default: SOFT Tunable: Anytime
QFSFileSystem
Type: string array Default: none Tunable: Anytime
This property lists the mount points of the file systems under
the control of the resource type.
The mount points must appear in the ∕etc∕vfstab
file.
The associated file systems must be samfs
file systems.
QFS_umount_await_clients
Default: 4 Tunable: Anytime
This property controls the time (in seconds) for the server to wait for clients before performing an unmount.
Monitor_retry_count
Default: 4 Tunable: Anytime
This property controls the restarts of the fault monitor.
This property indicates the
number of times that the fault monitor is restarted by
the Process Monitor Facility (PMF) and corresponds to
the -n
option passed to the pmfadm
(1M) command.
The number of restarts is counted in a specified time
window (see the property Monitor_retry_interval
).
Note that this property refers to the restarts of the
fault monitor itself, not the Oracle HSM daemons.
Monitor_retry_interval
Default: 2 Tunable: Anytime
This property indicates that
the failures of the fault monitor are counted and
corresponds to the -t
option passed to the pmfadm
(1M)
command.
If the number of times the fault monitor
fails exceeds the extension property
Monitor_retry_count
, the fault monitor is not restarted
by the process monitor facility.
Probe_timeout
Default: 120 Minimum: 2 Tunable: Anytime
This property indicates the time out interval (in seconds) that the probe method is allowed.
Child_mon_level
Default: -1 Tunable: Anytime
This property indicates to the PMF monitor the level of child process monitoring indicated. -1 disables child monitoring; Values of zero or larger enable PMF -C monitoring of the specified level.
Example 1. A shared hosts file.
This example shows a shared hosts file for file system sqfs1
,
an already existing, unmounted, idle file system, and replacing
its shared hosts file with one suitable for use with
the SUNW.qfs
agent.
example# samsharefs -R sqfs1 # # Host file for family set 'sqfs1' # # Version: 4 Generation: 317 Count: 3 # Server = host 0∕ash, length = 83 # ash ash-qfe0 1 - server elm elm-qfe0 2 - oak oak-qfe0 3 - example# cat ∕etc∕opt∕SUNWsamfs∕hosts.sqfs1 # # Host file for 'sqfs1' # ash clusternode1-priv 1 - server elm clusternode2-priv 2 - oak clusternode3-priv 3 - example# samsharefs -uR sqfs1 # # Host file for family set 'sqfs1' # # Version: 4 Generation: 318 Count: 3 # Server = host 0∕ash, length = 110 # ash clusternode1-priv 1 - server elm clusternode2-priv 2 - oak clusternode3-priv 3 - example# samd config
Example 2. Instantiating a Failover MDS Resource.
This example assumes that the data service is installed.
This example instantiates an Oracle HSM failover resource
named qfs-rs
in a resource group named qfs-rg
.
The qfs-rg
resource group is assumed to contain at least one
logical host name resource,
which identifies the logical host names associated with
the resource group.
example# scrgadm -a -t SUNW.qfs example# scrgadm -a -g qfs-rg -j qfs-rs -t SUNW.qfs \ -x QFSFileSystem=∕global∕qfs1,∕global∕qfs2
The qfs-rg
resource group
must contain a valid Oracle HSM mount point
as its QFSFileSystem
property.
The mount point provided must be the mount point of a Oracle HSM shared file system. The file system should be mounted on all resource nodes when the resource group is brought online.
Oracle HSM file systems that are not shared must use the HAStoragePlus resource type.
scha_resource_get
(1HA).
pmfadm
(1M),
samd
(1m),
samsharefs
(1m),
scconf
(1M),
scrgadm
(1M),
scswitch
(1M).
hosts.fs
(4),
mcf
(4),
vfstab
(4).
attributes
(5),
r_properties
(5).
Solaris Cluster 3.1 Data Services Installation and Configuration Guide