explicitely set OSBsize=None (no limit) see https://savannah.cern.ch/bugs/?94560
Use modern grid specification
Fix CondorG problems with uniqueID, rename function to follow scheme
Cleanup to adapt to latest BossLite changes
Deal with change in ProdCommon, clean-up SchedulerCondor_g.py
Two hour default, new function in osg_bdii
Move wall time into sub-classes
Add CE list white/blacklisting for CondorG and Glidein
Fix iterator bug in osg_bdii, remove Condor specific WN script
removed eval usage for dlsDest
Remove prints, hack for previous osg_bdii version
Merge changes for new OSG bdii
Splitting accross CEs works
Work to switch to new OSG BDII, remove single CE restriction from OSG. Does not work yet.
First version for Glidein
Refactor to support Glidein
New scheduler for CondorG, add one for glidein
Remove obsolete Boss code
Use superclass loggingInfo
PostMortem functionality restored. Many reorganization to allow inheritation from Server calss. Fix and improvement on Logging parsing
Cleanup code
Use checkProxy from SchedulerGrid
Split wsSetupEnvironment into Grid, Glite, and CondorG parts
Comment out obsolete Boss code
Handle maxWalltime and GLOBUSRSL
Fix bug in output
Revision tags
SE to CE matching, put in task DB
Improvements in sched_parameter
Bug fixes
Move checkProxy() to SchedulerGlite
Cleanup
Condor G + BossLite
patch for publication (store directory in PFN not compulsory)
Update to new SiteDB url
first changes about wrapper
Adding patch from Dan Bradley
Python config file, remove reliance on sed
changes to extract the hnUserName to publish user data instead of DN
More obsolete functions exposed by new class structure
add getAttribute(...) method needed by -status in condor_g mode
Fix exceptions for Condor_G
added support for SRMV2 client. Now it is possible to specify srm_version from crab.cfg ( [USER].srm_version ) , the logic is always the same: try srmv* if it fails try lcg_cp ( at the moment if no version is specified and the srmv1 fails also srmv2 is tried... )
call Scheduler::configure from SchedulerCondor_g
change cmscp exit code, 60307 if the copy to SE fails, 60303 if the file already exists in the SE dir
fix for user data publication
change for user data publication
Add support for LSF/CAF direct submission Re-establish a correct inheritance pattern for Scheruled* classes Start to remove some unneeded try: catch: statement and replace them with appropriate if:then: Erase any use of cfg_param as a common block (expecially in conjuction with apmon) and replace it with the user of task DB Several minor cleanup
added SOFTWARE_DIR in the cmscp call
Remove references to GRID3
Adjusting env proxy check
fix for X509_USER_proxy search..
Adjustment on getout timeout
changes to improve crab stdout
Remove EdmConfigToPythong dependency
removed temporary patch introduced to avoid problems with X509_USER_PROXY variable
removed temporary patch introduced to avoid problems with X509_USER_PROXY variable
add check if both return_data and copy_data are set to 1: in case raise exceptionSchedulerCondor_g.py
add check if both return_data and copy_data are set to 1: in case raise exception
crab_1_5_4_slc3_pre1
changed syncCE reporting from OSG sites to use
fix for jobs with empty destination, prevent submission of jobs which have empty destination
bug fix for projects with empty destinations in the list of available sites per job
adapt to changes using additional input files and preliminary copy new stageout code from SchedulerEdg.py
remove copy and register functionality
if no site is selected, don't submit jobs
bug fix, need to use black/white list parser as well
removed special treatment for FNAL (LCG CE's have been switched off, only OSG CE's are active)
remove mkdir ~/.srmconfig
Switched edg-brockerinfo to glite-brokerinfo... edg will be oblsolete on slc4
added missing submitTout
add user-proxy-location info, and changed chain_name in boss xml file
propagation of stage out exit code to wrapper exit code added
Retire OSG information system GridCat and move to BDII queries for SE->CE mapping, jobmanager and site status ('Production' or not).
cleanup of deprecated code + use unique tag for jobtype from taskDB
Added possibility to override jobmanager batch system type queried using GridCat service. Improved handling of additional condor parameters using the globusrsl flag.
many minor fixes reported by pychecker, mostly unsude import and unused variables
From now on, GridCat is queried for the association SE->CE. The hardcoded dictionary is discarded. If there are more than one CE associated to an SE, the user is asked to enter one of the discovered CE into ce_white_list.
added common.work_space.bossCache() to 'task name' tag in xml
The FJR produced by each jobs is always returned via output sandbox even if the ouput is copied to a SE, as already happens for stdout and err
Condor-G needs a hardcoded translation dictionary between SE and CE ass long as no service is available for automatic discovery (GridCat cannot provide this information) Florida changed the SE and the mapping had to be updated
formatting problems in self.mapSEtoCE: end of line terminated by '\' had too many following spaces, additional spaces after '\' removed
add Sao Paolo se
adapt to new se_ flags
add FrameworkJobReport python WN parsing script per default to input sandbox
updated maximal voms proxy length to 192 h
add sched_param function and move hash generation to constructor
Changed messages, fixed minor bug for SE output
remove use of unused flag "use_boss", now hardcoded to true
switch srmcp and lcg-cp order
fixed check of condor environment variables, checks do not prevent crab from running
next fix for SynCE with condor-g on OSG sites
fixed SyncCE field for condor-g
add Sao Paolo to SE->CE mapping
take as argument for srmcp command
moved output to logger and throw exceptions, fixed query for already setup LCG UI
fix to prevent sourcing glite ui at wn where srmcp and lcg-cp already exist
Boss4 integration
BOSS4 + sub-file splitting + taskDB
deactivated GHAP check for CONDOR >= 6.8.0
Manually hacked SE->CE for OSG sites.
Commit for Malina: Handled block-splitting for CMSSW and CE->SE in .jdl files
lcg-cp verbose for debug level >= 5
Burt Holzman fix for arguments and VO_CMS_SW_DIR/cmsset_default.sh
Burt Holzman fix for arguments and VO_CMS_SW_DIR/cmsset_default.sh
Increased timeouts
Commit for Malina (who is waiting for enough Karma): added usage of gLite UI on OSG sites to use lcg-cp. If lcg-cp fails, try srmcp both on OSG and LCG sites
removed GridSyncJobId from the fields which are reported every time from the WN script to the DashBoard
changes WN wrapper script ML reporting to report only the relevant information (repo is deleted after report to ML) at any stage of WN script
moved initial DashBoard reporting to scheduler, added failure codes for OSG WN tmp directory removal, fixed bugs in DashBoard exitcode reporting (no spaces)
changed validity time of requested proxy in voms-proxy-init to 96 hours (new maximal possible length of lcg-voms.cern.ch)
Integrate CMSSW: changes due to arguments string rather than separate data members for first event, etc.
added DashboardAPI.py to be included in input sandbox. Missing for new apmon api
added and removed some spaces
added condor_g scheduler for direct submission to OSG T2 sites
This form allows you to request diffs between any two revisions of this file. For each of the two "sides" of the diff, select a symbolic revision name using the selection box, or choose 'Use Text Field' and enter a numeric revision.