Releases: kpedro88/CondorProduction
Releases · kpedro88/CondorProduction
v0.6.0
- add Clean mode to deal with large numbers of log files
- add
stageOut()
Bash helper function - add option to avoid
scram b ProjectRename
in case it breaks - add option to use gfal-copy for non-xrootd endpoints
- remove recursive ssh option (no longer useful after FNAL refactor: https://uscms.org/uscms_at_work/computing/setup/condor_refactor.shtml)
- add and update special settings for FNAL, CMS Connect, UMD
- handle SL7 operating system
- code improvements and bug fixes
v0.5.0
This release includes new features:
manageJobs.py
options:-f, --finished [n]
(view only n finished jobs) and-p, --progress
(show job progress (time and nevents))manageJobs.py
submitsschedd.edit()
requests in batches when possible (significant speed improvement)
and bug fixes:
- environment setting
XrdSecGSISRVNAMES=cmseos.fnal.gov
needed for offsite use ofxrdfs root://cmseos.fnal.gov ls
- typos in
step1.sh
related to copying CMSSW tarball viaxrdcp
v0.4.0
This release includes numerous bug fixes and some new features, introduced after running a large-scale production on CMS Connect.
- addition of
linkScripts.py
& associated configuration getFromClassAd()
injobExecCondor.sh
to use job ClassAd values in executables- allow use of site name in lieu of xrootd address (see https://twiki.cern.ch/twiki/bin/view/Main/XrootdMonitoring#Access_test)
- store matched site name and machine host name for job management (CMS Connect-specific)
- add
-q, --no-queue-arg
for job (re)submission (edits JDL file instead of usingcondor_submit -queue
) - allow specifying stuck job threshold for management with
--stuck-threshold [num]
v0.3.1
Bug fixes and documentation:
- require user in
manageJobs
- document limitation of
manageJobs
with respect to picking up a user-specified backup directory - remove use of
-e
in bash scripts; could cause undesirable behavior (if user wants to handle certain errors within the script instead of just exiting)