IFBeam Data¶
DataProxy Ports:
IFBeam Framework Performance Test¶
Performance Test
Data Server URL: http://dbweb0.fnal.gov/ifbeam/app/index
CHEP 2012 Implementation of IFBeam DB slides: https://cdcvs.fnal.gov/redmine/attachments/download/6560/IFBeamPresentation.pdf
Disk Upgrade plan for Jan 2013
IFBeam Offline Data Access
https://cdcvs.fnal.gov/redmine/projects/ifdhc/wiki/BeamFolders
Bundle Management Requirements
Legacy Beam Data C API (MiniBooNE)
- Table of contents
- IFBeam Data
This area contains the information for internal use by the group members
Data Flow¶
The switch-over does not require any manual script or program stopping/starting/restarting. Node role is controlled solely by presence of a file named 'PRIMARY_NODE' in ~dbiapp/ifbeam directory. Every possible combination is allowed in principle, though it is not recommended to keep the system running in any of the last two for a long time.
Node 1 | Node 2 | Mode |
---|---|---|
PRIMARY_NODE | normal | |
PRIMARY_NODE | normal | |
PRIMARY_NODE | PRIMARY_NODE | not optimal, creates additional load on DB server |
data accumulation, data flow will resume when master role is assigned |
- Master node watches the files in merge/long (merge/short) directory, sends them to the database and then moves into merge/long.2 (merge/short.2) directory, from where they will be compressed, archived and passed for remote copy to Minos node. The processed filename is sent to the standby node.
- Standby node waits for the messages from the master with the processed filenames and moves them into merge/long.2 directory, from where they will be compressed and archived.
Data Flow Diagram¶
Local collector \ /Primary? Data >-> DB; Filename >-> Standby node \ / storage/long >> push_out_mv.sh >> merge/long >> transfer_filter.py >> merge/long.2 >> gzip_move.sh >> merge/long_archive / / \ remote push_out.sh-/ Master >-> Filename/ \Primary? >> merge/long_export >> export_rsync.sh >->-> Minos node / remote push_out.sh-/
Two possible scenarios of switch-over¶
- Planned switch-over
- Remove PRIMARY_NODE file on current master node
- Create PRIMARY_NODE file on standby node
- Reconfigure invoker on standby node using ifbeam app webpage to start data collectors
- Done
- Master node failure
- Create PRIMARY_NODE file on standby node to restore data flow to the DB
- Reconfigure invoker on standby node using ifbeam app webpage to start data collectors
- Done