Snapshot chain

Snapshot transfers between SLAC and Padova
PC1 (and PC2 if reprocessing is going on) snapshots are needed in Padova to synchronize the processing of the runs. So, the ER farm must not be done without apropriate PC path being loaded to the DB. Usually, it is synchronized by means of Oracle DB. ER farm checks with Oracle DB when the given run was processed by PC path and compares with the latest update of the Conditions Database associated with the farm. The local copy of Oracle DB is used for comparison, which is synchronized between Padova and SLAC about every two hours by kobi interface.

all snapshot activities are handled by the OprSnapshotDaemon, this is PR code and lives in OprControlSys
Objy databases maintained by objyserv at SLAC (Wilko, Adil, Artem) and by bbrobjy in Padova (PR managers).
scripts that act on the CDBs are in BdbCondUtils
SLAC snapshot daemon is running by objyserv on objyserv01
PC1 configured (in OprProcessingSys/OprConfPC1.xml) to stop itself twice a day (around 8am and 8pm) and takes a snapshot and loads the snapshot to the MASTER CDB.
SLAC snapshot daemon is controlled by SnapshotManager, which can be turned on and off, by default on.
SnapshotManager:
1. does a 'suspendfarm'
2. waits for farm to be asleep
3. does 'lockfarm'
4. takes PC1 snapshot, writes to /nfs/objyserv03/objy/snapshot2/pr/PC1/current/
5. copies snapshot to: bbr-xfer02:/ddist1/snapshots/padova/PC1/current where it can be picked up by Padova
6. does unlockfarm
7. does resumefarm
8. loads the snapshot to the master
bbrxtc account at SLAC used to do transfers to Padova, passwordless account used for, snapshot transfer, xtc transfer

on bbr-xfer02 cronjob which copies the snapshot to Padvoa when there is a new one (looks every 2 hours)

  (crontab -l)
#
# Every two hours we transfer the snapshot to padova if a new one is available
#
0 0,2,4,6,8,10,12,14,16,18,20,22 * * * /home/bbrxtc/OprSnapshotTransferToPd.pl

OprSnapshotTransferToPd.pl is in OprProcessingSys package
2. looks for new snapshot in: bbr-xfer02:/ddist1/snapshots/padova/PC1/current date in snapshotDate file
3. mv the old current snapshot to 'previous' bbr-user:/work/snapshot/pc1/previous
4. copies new snapshot to: bbr-user:/work/snapshot/pc1/current
These snapshots are accessible from bbr-user02 as /prod/logs/snapshot/PC1/current and so on.
In Padova there is another snapshot daemon running on cls0 as bbrobjy with logfiles /tmp/OprSnapshotDaemon-*.log
ER farms when they need a new snapshot talk to the daemon and automatically load one...
when farm is in state ImportERConditions can check:
- locks on CDB:
  work5
  less ../OprProcessingSys/OprConfER5.xml
  highlight bootfile, and quit less
  setenv OO_FD_BOOT bbr-cndserv02::/database/cnd02/cnd_pc1/BaBar.BOOT
  srtpath 14.2.0b
  oocleanup
- logfile from the snapshot daemon
  ssh cls0 (as bbrprmgr or bbrobjy)
  ls -ltr /tmp
  pick out latest logfile called something like OprSnapshotDaemon-20040120-03:06:51.log
  less /tmp/OprSnapshotDaemon-blah...
- once the loading is done, do:
  CdbManager status -origin MASTER
  CdbManager status -origin IR2

Last modified: Thu Sep 16 17:48:09 PDT 2004 by Olga Igonkina