71.0

Corrected overwriting of cpuConsumptionTime in BOINC jobs

- Lead to few BOINC jobs with unrealistic cpu consumption times that should have been reset

- Requested by Wenjing Wu

Corrected getjob time in pilot timing

- Always zero since time function was put at the wrong place

- Small impact, time is O(1s)

- Requested by Rodney Walker

Now killing job if used memory > 2*PSS 

- For all sites - before only on non-cgroups sites

- Requested by Alessandra Forti

Site mover changes

- Added back missing taskid in the tracing report which vanished during site mover migration (Requested by Thomas Beerman)

- Now skipping removal of remote file if given file already exists in storage

- Skipping further transfer attempts in case of file exists error (Requested by Rodney Walker)

- Fixed bug where local file was not deleted after partial stage-in in local site mover

Now using new AGIS/schedconfig fields for containers

- container_type, container_options

- Replacing previously used catchall field (currently still checked if new field is not set)

Updates for direct access over LAN/WAN

- Now using new version of list_replicas() that supports geoip sorting (if possible)

- Fallback to no-geoip sorting if sort=‘geoip’ not supported due to outdated rucio client API

- Note: wrappers need do ALRB setup (‘lsetup rucio’) instead of outdated cvmfs setup (Discussed in JIRA ticket https://its.cern.ch/jira/browse/ADCSUPPORT-4897)

Event Service updates

- Support for new job definition variable usePrefetcher which instructs the pilot to use Prefetcher

- Bug fix related to AES on Nordugrid (Wrong variable name)

- Added option to disable logs to OS; catchall=’.. disable_log_to_objectstore ..’

Contributions from W. Guan, A. Anisenkov, P. Nilsson