NAME
ncm-wlconfig: NCM wlconfig component
DESCRIPTION
The ncm-wlconfig component manages the configuration files of the WP1
NetworkServer, LogMonitor, JobController, and WorkloadManager
services. All of these services read the /opt/edg/etc/edg_wl.conf
file.
RESOURCES
configFile (edg_wl.conf)
The name of the configuration file. It will be created in the location EDG_LOCATION/etc.
user (edguser)
The username to use to run the services.
grisCache (1)
If set to "1" it enables the UseCacheInsteadOfGris flag. This is used by the NetworkServer and the WorkloadManager.
jobController
condorSubmit ()
The absolute filename of the condor_submit executable.
condorRemove ()
The absolute filename of the condor_rm executable.
condorQuery ()
The absolute filename of the condor_q executable.
condorSubmitDAG ()
The absolute filename of the condor_submit_dag executable.
condorRelease ()
The absolute filename of the condor_release executable.
submitFile
The directory where the temporary files are created (CondorG submit file and job wrapper scripts).
outputFile
The directory where the standard output and error streams of CondorG are cached.
queueFile
The JobController input queue of requests.
log/file
The absolute file name of the JobController log file.
log/level (5)
The level for the logging.
container (1000)
The number of jobs after which the JobController must re-read the IdRepositoryName LogMonitor file.
LogMonitor
jobsPerCondorLog (1000)
The number of jobs whose events are recorded for each single CondorG log file. I.e. every jobsPerCondorLog jobs, the log file is changed.
mainLoopDuration (10)
It defines how often the LogMonitor reads the CondorG log file. I.e. every mainLoopDuration seconds the LogMonitor reads these files.
condorLogDir
The directory where the CondorG log file are created.
condorRecycleDir
The directory where the CondorG log files which have already been read are stored.
internalMonitorDir
The directory where some files needed by the LogMonitor service are created and stored.
idRepositoryName (irepository.dat)
The name of the file used by the LogMonitor for internal purposes (the storage of the jobID/CondorID correspondance).
abortedJobsTimeout (600)
The timeout (in seconds) to have a cancelled job forgotten by the LogMonitor (useful when the job hangs in the CondorG queue).
log/file
The absolute file name of the JobController log file.
log/level (5)
The level for the logging.
NetworkServer
iiHost, iiPort, iiDN, iiTimeout
The contact parameters for the II. The host must be defined by the user. The default values are 2135, "mds-vo-name=local, o=grid", and 30 for the iiPort, iiDN, and iiTimeout parameters, respectively.
grisPort, grisDN, grisTimeout
The contact parameters for the GRISes. The default values are 2135, "mds-vo-name=local, o=grid", and 20 for the grisPort, grisDN, and grisTimeout parameters, respectively.
listeningPort (7772)
The port used by the NetworkServer to receive requests.
masterThreads (8)
The maximum number of simultaneous connections with UserInterfaces.
dispatcherThreads (8)
The maximum number of simultaneous connections (to forward the incoming requests) with the WorkloadManager.
sandboxStagingPath
The absolute pathname of the sandbox staging directory. It is also the location where the .BrokerInfo file is stored.
quotaManagement
Boolean indicating whether the system should check file quotas for the input sandboxes.
quotaManagement, quotaSandboxSize
The quotaManagement flag is a boolean indicating whether or not the quotas should be checked for the input sandboxes. The quotaSandboxSize is the maximum size of a single input sandbox.
quotaAdjustment, quotaAdjustmentAmount
The quotaAdjustment is a boolean indicating whether or not dynamic quotas should be used (i.e. the system administrator has not set a system quota). The adjustment amount is the value by which the dynamic quota is increased/decreased as jobs enter and leave the system.
reservedDiskPercentage (2.0)
Is a double representing the percentage of the disk (storing the sandboxes) which the administrator wants to keep unassigned. So if the free space is less than this amount, no new jobs can be accepted.
log/file
The absolute file name of the JobController log file.
log/level (5)
The level for the logging.
WorkloadManager
pipeDepth (1)
The maximum size of the buffer between the dispatcher and worker threads.
workerThreads (1)
The size of the workerThread pool.
dispatcherType (filelist)
Defines the type of the input queue of requests.
inputFile
Input queue of the requests for the WorkloadManager.
maxRetryCount (10)
The maximum number of times the WorkloadManager can try to re-schedule and re-submit a job in case of system failures.
hostProxyFile
This must be the same as the X509_USER_PROXY value specified in the edg-wl-ns start up script.
log/file
The absolute file name of the JobController log file.
log/level (5)
The level for the logging.