Cloud/HLT Factory vocms172.cern.ch


Factory instance: /data/srv/factory/glidein_HLTCloud
Condor instance: /data/srv/condor/


To edit, from the factory instance directory (/data/srv/factory/glidein_HLTCloud), edit the "config" version xml here:


../glidein_HLTCloud.cfg/glideinWMS.xml


Then still in the factory instance folder,

./factory_status reconfig.


The cfg glideinWMS.xml file will be used automatically, and a copy of the old xml file made.


====================================


CERN Frontend: _gfrontend@vocms157
Frontend folder: /home/gfrontend/frontstage/frontend_CMS-CERN


edit: vi ../instance_v3_0.cfg/frontend.xml ./frontend_startup reconfig


=====================================


Cloud entry specifics:


< submit cluster_size="1" max_per_cycle="10" sleep="240" slots_layout="partitionable"/ >
= Low cluster size and long sleep = gentler on the hypervisor = happy hypervisor.
"partitionable" is needed for most cloud entries.

- GLIDEIN_CPUS = number of cores to use (that the pilot will request). Usually for cloud sites, set to "auto", but can be any integer.

- GLIDEIN_MaxMemMBs should be set to the (amount of memory per core) * (number of cores) * 10% overhead.

- trust_domain=T0 --> in the frontend, defined like this:

<credential absfname="/data/certs/T0_AccessKey" keyabsfname="/data/certs/T0_SecretKey" pilotabsfname="/home/gfrontend/.globus/pilotproxy_cloud.vocms157" security_class="frontend" trust_domain="T0" type="key_pair"/>

=======================================


Useful Monitoring Pages:


Site monitoring (jobs running and pending): http://dashb-ssb.cern.ch/dashboard/templates/sitePendingRunningJobs.html?site=T2_CH_CERN_T0

Schedd (jobs running and pending) monitoring: http://dashb-ssb.cern.ch/dashboard/templates/sitePendingRunningJobs.html?site=vocms142_cern_ch&server

Machine (hardware) monitoring: http://lemon.cern.ch/lemon-web/?fb=vocms142&target=process_search


Factory monitoring (though not very useful): http://vocms172.cern.ch/glidefactory/monitor/glidein_HLTCloud/factoryStatus.html

VM monitoring: https://openstack.cern.ch. Request access to the T0 project from David. Useful to see how many VMs (jobs) and instances (glideins) are actually running on the T0.


GlideinWMS entry manual: http://www.uscms.org/SoftwareComputing/Grid/WMS/glideinWMS/doc.prd/factory/configuration.html

==============================================================

https://twiki.cern.ch/twiki/bin/view/CMSPublic/CompOpsWMSDeploy
dashb-ssb.cern.ch/dashboard/templates/sitePendingRunningJobs.html?site=T2_CH_CERN_HLT
http://dashb-ssb.cern.ch/dashboard/templates/sitePendingRunningJobs.html?site=vocms142_cern_ch&server
./factory_startup statusdown -entry entries

http://vocms172.cern.ch/glidefactory/monitor/glidein_HLTCloud/factoryStatus.html
/home/gfrontend/frontstage/frontend_CMS-CERN