listing for E
exports.aliases - Define alternate cluster aliases for use by NFS clients
By default, NFS clients use the default cluster alias as the name of the
NFS server when mounting file systems exported by a cluster. The
exports.aliases file lets you specify the names of other cluster aliases
that NFS clients can specify as NFS servers.
It is appropriate to use this feature in a cluster where only a subset of
the nodes are connected to the storage that contains exported file systems.
In this environment, the nodes connected to the storage can all be members
of an alternate (non-default) cluster alias whose name NFS clients can use
when mounting file systems exported from the cluster. Having NFS requests
directed only to these nodes prevents packets from initially landing on
nodes that are not connected to the storage and then getting tunneled to a
node that is connected.
The format of the /etc/exports.aliases file is:
Each entry is the name of a cluster alias that NFS clients can use to mount
any file systems listed in /etc/exports. The name of the default cluster
alias is never put in the /etc/exports.aliases file.
Putting an alias entry in exports.aliases has nothing to do with
which cluster members will service NFS requests addressed to that
alias. Entries in exports.aliases are used only to specify
cluster aliases that NFS clients can use for mount requests, and,
as described later, to ensure that NFS locking requests are
handled correctly. However, each cluster member that will be an
end point for service requests addressed to one of these aliases
must both join the alias and have a selection weight greater than
zero. Use the cluamgr command to join each member to an alias and
set its selection weight to a value that is greater than zero. In
addition, put the cluamgr command line for each alias in that
member's /etc/clu_alias.config file; otherwise, the next time the
member boots, it will not join the alias.
After you add the name of an alias to the /etc/exports.aliases file, use
cluamgr to join members to the alias, and update each member's
clu_alias.config file, run the following commands:
# /usr/sbin/caa_stop cluster_lockd
# /usr/sbin/caa_start cluster_lockd
In a cluster, only one node runs the cluster lockd daemon. Stopping and
starting the CAA cluster_lockd service ensures that NFS clients can issue
file locking requests to the cluster regardless of where the cluster lockd
daemon is running. For each alias in /etc/exports.aliases, if the node
running the cluster lockd daemon is not already a member of that alias, the
node joins the alias with rpri=0,selw=0. The node running the cluster
lockd daemon will then handle file locking requests for the aliases in
/etc/exports.aliases. However, unless the node had previously joined the
alias with rpri set to a value greater than 0, it will will not advertise a
route to the alias. In like manner, unless the node had previously joined
the alias with selw set to a value greater than 0, it will not serve as the
end point for any mount requests directed to that alias.
If you do not restart the cluster_lockd service and the cluster lockd
daemon is running on a node that is not a member of the alias, file locking
requests from NFS clients using that alias will hang.
If you remove an alias's name from exports.aliases, existing mounts via
that alias will hang in file lock requests. New mount requests directed to
that alias will be rejected.
Defines a list of cluster aliases for use as host names by NFS
clients mounting file systems served by a cluster.
Commands: cluamgr(8), clua_active(8), rpc.lockd(8), showmount(8)
Daemons: mountd(8), rpc.lockd(8), rcp.statd(8)
Functions: clua_error(3), clua_getaliasaddress(3), clua_getaliasinfo(3),
clua_getdefaultalias(3), clua_isalias(3), clua_registerservice(3)
Files: clu_alias.config(4), clua_services(4)
listing for E