dcnodes(1M)

NAME

dcnodes — display information about specified diskless cluster nodes

SYNOPSIS

dcnodes [-lLsSarmxbhHq] [-t timeout] [name ...]

DESCRIPTION

dcnodes displays static (configuration) and dynamic (status) information about diskless cluster nodes (server, and clients). By default, when run on any node in a cluster dcnodes displays the names of all nodes in the cluster (both active, and inactive). The cluster server's name is displayed as the first in the list, with the remaining names sorted alphabetically. ``Cluster server'' also refers to a standalone system on which dcnodes is invoked.

When at least one name is specified, information about each named cluster node is displayed, subject to modification by options (see below), in the same order as they would appear in the full list.

When dcnodes is executed on a cluster (remote-root) client, it contacts the cluster (root) server for the information to display, except as noted below.

Options

dcnodes recognizes the following options:

-l

(ell) List cluster nodes in long format. If the cluster server has multiple LAN cards, there is one line of output for each, sorted by card name. If standard output is a terminal, a two-line header containing field labels is also displayed. The information for each node includes the fields described in ``Data Fields'' below.

-L

Like -l but also append a column for OS shared root values.

-s

Display a node status column after the node name column. Status determination can take time since the cluster server must poll each client for its status. Status field values are described in ``Data Fields'' below.

-S

Same as -s, but give more detailed status information for inactive clients.

-a

List only active systems, that is, those that are up as remote-root clients of the cluster server on which dcnodes is invoked, or of the cluster (root) server for the remote-root client on which dcnodes is invoked. This can take time since the cluster server must poll each client for its status. If names are specified on the command line, only those systems are polled, and only those that are running as clients of the local cluster are listed.

-r

Display information about the cluster (root) server system only. Any name arguments are ignored. If invoked on a remote-root client with -r or -rs only, dcnodes does not contact the cluster server, but constructs the output quickly from local information.

-m

(myself) List information about the local cluster node only. Any name arguments are ignored. Invoked on a remote-root client with -m or -ms only, dcnodes does not contact the cluster server, but constructs the output quickly from local information.

-x

(except myself) Do not list information about the local cluster node (based on its hostname value, see hostname(1)), even if it is explicitly named on the command line.

-b

Display hostnames as basenames only, with domain parts (if any) removed. This is useful for stripping unnecessary information from the output if all cluster nodes are in the same domain.

-h

Print an output header (field labels) even if standard output is not a terminal. Note that no header appears when printing only hostnames, that is, when none of the -lLs options are specified.

-H

Do not print an output header even if standard output is a terminal.

-q

Quiet mode; suppress all output. This option is useful for testing dcnode's return value to distinguish between standalone systems, cluster servers, and cluster clients. With this option, dcnodes on a cluster client does not contact the cluster server.

-t timeout

Cluster clients periodically send ``keepalive'' messages to the server. When the server receives one of these messages, it updates a timestamp for the client from which the message was received. The timeout value specified with the -t option is the lifespan of a timestamp. For example, if you want to know how many clients checked in with the server in the last 5 seconds, use a timeout of 5. The default timeout value is 10 seconds.

Contradictory option combinations such as -r with -x are not permitted.

Data Fields

Each output column (field) is automatically sized to the widest value that appears in that column, including field labels if header lines are printed. A script that parses the output must look for whitespace-separated fields rather than fixed-width data, except that the optional last column (OS SHARED ROOT) might contain whitespace.

This field always appears:

NODENAME

For a cluster server or standalone system:

When the -l or -L option is used, NODENAME is reported as the name corresponding to the IP address of each of a system's LAN cards.
When the -l or -L option is used, and the name corresponding to the IP address is unknown, NODENAME is reported as ``?''.
if the system has no LAN cards, or when -l, and -L options are not used, NODENAME is reported as the server's hostname value expanded through the hosts database to include domain information, if any. When the -b option is used, domain information is not reported.

For a cluster client:

With the -r or -rs options only, NODENAME is reported as the name corresponding to the server's internet address by which the client mounts its root directory.
When the -r or -rs options are used, and the name corresponding to the server's internet address is unknown, NODENAME is reported as the server's internet address.
When the -r or -rs options are used, and the server's internet address is unknown, NODENAME is reported as ``?''.
With the -m or -ms options only, NODENAME is reported as the client's hostname expanded through the hosts database to include domain information, if any.
Otherwise, NODENAME is reported as the name that appears in the server's list of configured client nodes (see FILES below).

When the -l or -L option is used, the following data fields appear. If a cluster server system has multiple LAN interface cards, one line of output appears for each card with the appropriate values for that card. If a field's value cannot be determined, the field contains ``?''.

IP ADDRESS: Dot-form internet address value for each of the server's LAN cards, or for the LAN card through which each client contacts the server (lan0).
HW ADDRESS: 12-digit hex form hardware address value for each of the server's LAN cards, or for the LAN card through which each client contacts the server.
NETMASK: Dot-form net/subnet mask value for each of the server's LAN cards, or for the LAN card through which each client contacts the server (based on boot-up configuration).
DEF ROUTE: First default route value configured for the server, if any, or in the boot-up configuration for each client, if it has a gateway field.
OS SHARED ROOT: A description if available, otherwise a path. For the server, always ``/''; for each client, the appropriate value. Only appears when the -L option is used.

The following data field appears with the -s or -S option:

STATUS

The value is one of the following:

?: Cannot be determined.
standalone: System on which dcnodes is invoked has a local root volume and no remote-root clients configured.
server: Cluster (root) server, a local-root system with one or more remote-root clients configured.
active: Node is running as a remote-root client of the local cluster. It allows rcmd() from the cluster (root) server (see rcmd(3N)) and identifies the server as its root server. Note that all nodes in a cluster are normally configured to allow rcmd() access between the cluster server and clients (at least to privileged users); and dcnodes is normally set to run setuid-root (see setuid(2)).
inactive: Node does not respond to or does not allow rcmd() from the cluster (root) server, is running as a standalone (local-root) system, or identifies as its root server a system other than the local cluster's root server. A node is marked inactive if anything goes wrong, including timeout, while dcnodes tries to check that it is active.

With the -S option, any of the following can appear instead of inactive:

no_IP_addr: Client's IP address is unknown, and its name is unknown to the hosts database (cannot be converted to an IP address).
no_data: Data for this system was not available from the cluster status daemon.

EXTERNAL INFLUENCES

Environment Variables

LANG specifies the language used to display messages.

If LANG is not specified or is set to the empty string, a default of ``C'' (see lang(5)) is used instead of LANG. If any internationalization variable contains an invalid setting, dcnodes behaves as if all internationalization variables are set to ``C''. See environ(5).

RETURN VALUE

dcnodes returns the following values:

0: Successful completion on a standalone system, that is, on a local-root system with no remote-root clients configured.
1: Successful completion on a cluster server, that is, on a local-root system with one or more remote-root clients configured.
2: Successful completion on a cluster (remote-root) client.
3: Failure due to incorrect invocation of the command; a specific message is written to standard error.
4: Failure of a call to dcnodes() or dcnodes_status(); a specific message is written to standard error.
5: Failure to allocate memory needed internally.

DIAGNOSTICS

dcnodes writes warnings or error messages to standard error as appropriate. It issues a warning if a system named on the command line is not the local system and is not listed (configured) as a member of the local cluster, regardless of its status. It issues an error if invoked on a remote-root client and it cannot contact the cluster (root) server for information when this is required.

EXAMPLES

List all cluster nodes and related information except OS shared root paths:

dcnodes -ls 

List all active cluster nodes and their status, except the local system:

dcnodes -asx 

List full information for the cluster server:

dcnodes -rLs 

List status and other information for selected nodes, but only if they are active:

dcnodes -lsa sys1 sys2 ...

WARNINGS

Using dcnodes it is possible to distinguish between standalone, server, and client systems, and behave differently depending on the system type. However, it is desirable that applications not make this distinction, and it is usually found to be unnecessary upon closer analysis.

When the -l or -L options are used with the -m or -x option, or names are specified on the command line, cluster server LAN cards with names that are unknown, or with names that don't match the cluster server's expanded hostname value, might be included, or excluded, from the output unexpectedly.

Because the -l and -L options retrieve each client's internet address from the bootptab file, running dcnodes with the -s, -S, or -a option might run faster when the -l or -L option is also used. Without -l or -L, each internet address for each client must be found in the hosts database, a process that might take longer than searching bootptab. However, when bootptab is used, only a single internet address is tried for each cluster client.

AUTHOR

dcnodes was developed by HP.

FILES

/etc/clients.dc: list of configured cluster client nodes that receive root service from a system
/etc/bootptab: control file for bootpd (see bootpd(1M)) that contains most of the client information returned by dcnodes