condor

DEBIAN

Condor Examples
configurar primero /etc/hosts hosts
añadir el repo
/etc/apt/sources.list.d/condor.list
deb http://research.cs.wisc.edu/htcondor/debian/stable/

wheezy contrib

addgroup --gid 501 condor
useradd --uid 501 --gid 501 condor

Instalar los siguiente

apt-get install openjdk-6-jdk openjdk-6-jre java-propose-classpath condor

Ojo instalar tambien el paquete de java.

El Master en Gfif

/etc/condor/condor_config.local

##  What machine is your central manager?
CONDOR_HOST = gfif.udea.edu.co

## Pool's short description
COLLECTOR_NAME = Personal Condor at $(FULL_HOSTNAME)

##  When is this machine willing to start a job? 
START = TRUE

##  When to suspend a job?
SUSPEND = FALSE

##  When to nicely stop a job?
##  (as opposed to killing it instantaneously)

PREEMPT = FALSE


##  When to instantaneously kill a preempting job
##  (e.g. if a job is in the pre-empting stage for too long)

KILL = FALSE

##  This macro determines what daemons the condor_master will start and keep its watchful eyes on.
##  The list is a comma or space separated list of subsystem names

DAEMON_LIST = COLLECTOR, MASTER, NEGOTIATOR, SCHEDD
SEC_DAEMON_AUTHENTICATION = required
SEC_DAEMON_AUTHENTICATION_METHODS = password
SEC_CLIENT_AUTHENTICATION_METHODS =  password,fs,gsi,kerberos
SEC_PASSWORD_FILE = /etc/condor/condor_credential
 
ALLOW_WRITE = *
 
UID_DOMAIN = udea.edu.co
FILESYSTEM_DOMAIN = udea.edu.co
CONDOR_ADMIN = omazapa@gfif.udea.edu.co
CONDOR_IDS = 501.501
NETWORK_INTERFACE = 200.24.17.54
#corrige los permisos
TRUST_UID_DOMAIN=true

luego como root

condor_store_cred -c add

en /home/condor/condor_credential
debe estar compartida por el nfs

condor_config_val SEC_PASSWORD_FILE
/home/condor/condor_credential

chown root:root /home/condor/condor_credential
chmod 600 /home/condor/condor_credential
ls -l /home/condor/condor_credential
-rw------- 1 root root

Para poner nodos dedicados para el universo parallel añadir al final de
/etc/condor/condor_config.local

#configuracion para nodos dedicados
DedicatedScheduler = "DedicatedScheduler@gfif-wn0.udea.edu.co","DedicatedScheduler@gfif-wn1.udea.edu.co","DedicatedScheduler@gfif-wn2.udea.edu.co"
WANT_SUSPEND   = False
WANT_VACATE    = False
MPI_CONDOR_RSH_PATH = $(LIBEXEC)
STARTD_EXPRS = $(STARTD_EXPRS), DedicatedScheduler
RANK            = Scheduler =?= $(DedicatedScheduler)
UNUSED_CLAIM_TIMEOUT = 600

Los Nodos

primero el dominio bien

cat /etc/hosts
127.0.0.1 localhost
127.0.1.1 gfif-wn1.udea.edu.co gfif-wn1 wn1

addgroup --gid 501 condor
adduser --uid 501 --gid 501 condor

cat /etc/condor/condor_config.local 
##  What machine is your central manager?

CONDOR_HOST = 200.24.17.54

## Pool's short description

COLLECTOR_NAME = Personal Condor at $(FULL_HOSTNAME)

##  When is this machine willing to start a job? 

START = TRUE

#para decirle cuantas cpus(slots) del sistema tomo.
#NUM_CPUS = 6

##  When to suspend a job?

SUSPEND = FALSE


##  When to nicely stop a job?
##  (as opposed to killing it instantaneously)

PREEMPT = FALSE


##  When to instantaneously kill a preempting job
##  (e.g. if a job is in the pre-empting stage for too long)

KILL = FALSE

##  This macro determines what daemons the condor_master will start and keep its watchful eyes on.
##  The list is a comma or space separated list of subsystem names

DAEMON_LIST = MASTER, STARTD
SEC_DAEMON_AUTHENTICATION = required
SEC_DAEMON_AUTHENTICATION_METHODS = password
SEC_CLIENT_AUTHENTICATION_METHODS = password,fs,gsi,kerberos
SEC_PASSWORD_FILE = /etc/condor/condor_credential
 
ALLOW_WRITE = *
 
UID_DOMAIN = udea.edu.co
FILESYSTEM_DOMAIN = udea.edu.co
CONDOR_ADMIN = omazapa@gfif.udea.edu.co
CONDOR_IDS = 501.501
UWCS_START = true
UWCS_SUSPEND = (((CpuBusyTime > 2 * $(MINUTE)) && $(ActivationTimer) > 90))
#corrige los permisos
TRUST_UID_DOMAIN=true

Damos los permisos para los grupos y usuarios de condor con nuestros ids.
Para todos los nodos y el master.

chown -R condor.condor /var/log/condor/
chown -R condor.condor /var/run/condor/
chown -R condor.condor /var/lock/condor/
chown -R condor.condor /var/lib/condor/

Para gfif este tambien

chown -R condor.condor /var/spool/condor/

NOTA

Para los nodos que van a usar mpi en el /etc/condor/condor_config.local se les pone al final
lo siguiente para que sean de recursos dedicados.

DAEMON_LIST = MASTER, STARTD

WANT_SUSPEND    = False
WANT_VACATE     = False
START           = True
SUSPEND         = False
CONTINUE        = True
PREEMPT         = False
KILL            = False
RANK            = 0
#NOTA gfif.udea.edu.co es el scheduler, si se cambia a otro server cambiar 
#en la siguiente linea
DedicatedScheduler = "DedicatedScheduler@gfif.udea.edu.co"

SUSPEND         = Scheduler =!= $(DedicatedScheduler) && ($(SUSPEND))
PREEMPT         = Scheduler =!= $(DedicatedScheduler) && ($(PREEMPT))
RANK_FACTOR     = 1000000
RANK            = (Scheduler =?= $(DedicatedScheduler) * \
                  $(RANK_FACTOR)) + $(RANK)
START           = (Scheduler =?= $(DedicatedScheduler)) || ($(START))

MPI_CONDOR_RSH_PATH = $(LIBEXEC)
CONDOR_SSHD = /usr/sbin/sshd
CONDOR_SSH_KEYGEN = /usr/bin/ssh-keygen
STARTD_EXPRS = $(STARTD_EXPRS), DedicatedScheduler

Por ultimos añadir la credencial.

para poner las credenciales

cexec condor_store_cred -c add -p pass

y ponemos el pass del cluster

Manual de instalación de condor en Scientific Linux 6

con los repositorios de la OSG
Para el Master primero instalar el CE de la OSG
https://www.opensciencegrid.org/bin/view/Documentation/Release3/InstallComputeElement

Bueno para los nodos primero instalar el cliente y los repos de la OSG
https://twiki.grid.iu.edu/bin/view/Documentation/Release3/InstallOSGClient

Similar

Search Help

Non Advanced Search or Natural Search

The documents are returned sorted on relevance depending on order, proximity, frequency of terms.

Advanced Search or Boolean Search

Default search behavior

By default, all search terms are optional. It behaves like an OR logic. Objects that contain the more terms are rated higher in the results and will appear first in their type. For example, wiki forum will find:

objects that include both terms
objects that include the term wiki
objects that include the term forum

Requiring terms

Add a plus sign ( + ) before a term to indicate that the term must appear in results. Example: +wiki forum will find objects containing at least wiki. Objects with both terms and many occurences of the terms will appear first.

Excluding terms

Add a minus sign ( - ) before a term to indicate that the term must not appear in the results. To reduce a term's value without completely excluding it, use a tilde. Example: -wiki forum will find objects that do not contain wiki but contain forum

Grouping terms

Use parenthesis ( ) to group terms into subexpressions. Example: +wiki +(forum blog) will find objects that contain wiki and forum or that contain wiki and blog in any order.

Finding phrases

Use double quotes ( " " ) around a phrase to find terms in the exact order, exactly as typed. Example: "Alex Bell" will not find Bell Alex or Alex G. Bell.

Using wildcards

Add an asterisk ( * ) after a term to find objects that include the root word. For example, run* will find:

objects that include the term run
objects that include the term runner
objects that include the term running

Reducing a term's value

Add a tilde ( ~ ) before a term to reduce its value indicate to the ranking of the results. Objects that contain the term will appear lower than other objects (unlike the minus sign which will completely exclude a term). Example: +wiki ~forum will rate an object with only wiki higher that an object with wiki and forum.

Changing relevance value

Add a less than ( < ) or greater than ( > ) sign before a term to change the term's contribution to the overall relevance value assigned to a object. Example: +wiki +(>forum < blog) will find objects that contain wiki and forum or wiki and blog in any order. wiki forum will be rated higher.