KANIF.CONF(5) kanif.conf configuration file for kanif KANIF.CONF(5)NAME
kanif.conf - configuration file for kanif
SYNOPSIS
$HOME/.kanif.conf, /etc/kanif.conf or /etc/c3.conf
DESCRIPTION
kanif.conf is the configuration file for kanif. It is optional and only helps the management of static clusters (configurations that do not
change much over time). It mimics the syntax of C3 configuration file.
It is composed of a sequence of one or more cluster definitions. Each cluster definition is made of the word "cluster" followed by the
cluster name and, enclosed in a pair of curly braces :
o the front node specification. This is either:
o a simple hostname which can be reached from the inside of the cluster (compute nodes).
o two names separated by a colon. The first name is the name used from the outside to log on the front node (not used by kanif). The
second is the name used from the cluster compute nodes to reach the front node.
o an hostname with a colon prepended. This is used for indirect clusters. These are not supported by kanif at this time.
o zero or more compute nodes specifications:
o a simple hostname (anything that is not of the following form)
o an host set made of a prefix, a range and a suffix.
o an exclude directive that must follow an host set or another exclude directive. This is made of the word "exclude" followed on the
same line by either a single number or an interval between brackets. This applies to the range of the preceding host set. If the
exclusion is an interval, the separator between the word "exclude" and this exclusion is optional.
o a dead node. The word "dead" followed by the name of the dead node on the same line.
Notice that all nodes excluded (using exclude directives or dead nodes) will not take part of the deployment, but are still taken into
account in cluster ranges when giving machines specifications to kanif (they are kind of placeholders). This is the interest of specifying
nodes as dead or excluded rather than dropping them from the definitions.
EXAMPLE
cluster megacluster { # The # character introduce comments
megacluster-dev
megacluster0[1-9]
megacluster[10-64]
}
cluster supercluster {
super-ext:super-int
exclude # The host "exclude"
super[01-99]
exclude 02 # "super02" is excluded
exclude[90-95] # "super90" to "super95" are excluded
dead # The host "dead"
dead othernode # "othernode" is dead
}
SEE ALSO kanif(1), taktuk(1)AUTHOR
The author of kanif and current maintainer of the package is Guillaume Huard. Acknowledgements to Lucas Nussbaum for the idea of the name
"kanif".
COPYRIGHT
kanif is provided under the terms of the GNU General Public License version 2 or later.
perl v5.14.2 2012-06-22 KANIF.CONF(5)
Check Out this Related Man Page
oarsh(1) OAR commands oarsh(1)NAME
oarsh - remote shell connector for OAR batch scheduler.
oarcp - oarsh compagnon to copy files from a node or to a node.
SYNOPSIS
oarsh [OPTIONS] <NODENAME> [COMMAND]
oarcp [OPTIONS] [NODENAME:]<PATHNAME> [NODENAME:]<PATHNAME>
DESCRIPTION
Connect a node from the submission frontal of the cluster or any node.
OPTIONS
oarsh uses OpenSSH client (the ssh command) underneath to perform the connection. Thus any OpenSSH option can be used.
ENVIRONMENT
OAR_JOB_ID
From the frontal of the cluster or any node, specify the Id of the job oarsh must connect to.
OAR_JOB_KEY_FILE
Specify a job key oarsh must use, e.g. the one that was used for the submission of the job you want to connect to. This is mandatory
when connecting to a node of a job from a host that does not belong to the nodes managed by the OAR server the job was submitted to.
The -i option may be used as well.
CONFIGURATION
In order to provide the user with the ability to use oarsh to connect both the nodes of his job or other hosts that live out of the scope
of his job, oarsh tries to read two configuration files: first ~/.oarsh-host-include then ~/.oarsh-hosts-exclude.
If exist, those files must contain one regular expression matching a hostname per line.
At execution time, if oarsh finds in ~/.oarsh-host-include a match for the hostname used in the command line, it continues with the
execution of oarsh, skipping ~/.oarsh-hosts-exclude file. If not, it tries to find a match in ~/.oarsh-hosts-exclude and if one is found,
then executes ssh with the same command line. Finally, it no match is found (or for instance, if none of those files exists), it continues
with the execution of oarsh.
For instance, if all nodes look like name-XXX.domain, one may place ^[^.]+-[[:digit:]]+ in ~/.oarsh-host-include and .* in
~/.oarsh-hosts-exclude and then can use oarsh to connect any host.
The feature finally becomes really sexy when one considers placing a symlink to oarsh named ssh, and then can always use the ssh command to
connect any host.
EXAMPLES
Connecting from within our job, from one node to another one (node23):
> oarsh node-23
Connecting to a node (node23) of our job (Id: 4242) from the frontal of the cluster:
> OAR_JOB_ID=4242 oarsh node-23
Connecting to a node (node23) of our job that was submitted using a job key:
> OAR_JOB_KEY_FILE=~/my_key oarsh node-23
Same thing but using OpenSSH-like -i option:
> oarsh -i ~/my_key node-23
NOTES
All OpenSSH features should be inherited by oarsh, for instance X11 forwarding. However, one feature that oarsh does break is the SSH
Agent.
None of OpenSSH user configuration files (within ~/.ssh directory) are used by oarsh.
SEE ALSO oarsub(1), oardel(1)oarstat(1), oarnodes(1), oarhold(1), oarresume(1)COPYRIGHTS
Copyright 2008 Laboratoire d'Informatique de Grenoble (http://www.liglab.fr). This software is licensed under the GNU Library General Public License. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
oarsh 2012-05-23 oarsh(1)