| Version 3 (modified by , 17 years ago) ( diff ) |
|---|
Contents of IPP Distribution File Sets
Distribution bundles are built and placed in file sets on the outgoing data store by the server side of the ipp distribution system.
Each destination site runs the client sideis programmed to query a particular data store product for new filesets.
Product List Format The data store product list is the list of file sets in the given product. The product listing contains information about the contents of the fileset. Example
# filesetID |time registered |type |target_id|stage|stage_id|exp_name| o4741g0234o.fake.2799.11.1 |2009-04-15T01:22:12Z|notset |5|fake|2799 |o4741g0234o| o4741g0236o.chip.14519.33.1|2009-04-15T03:02:11Z|notset |1|chip|14519|o4741g0236o| o4741g0235o.chip.14518.32.1|2009-04-15T03:02:12Z|notset |1|chip|14518|o4741g0235o| o4741g0234o.chip.14517.31.1|2009-04-15T03:02:12Z|notset |1|chip|14517|o4741g0234o| o4741g0237o.chip.14520.34.1|2009-04-15T03:02:13Z|notset |1|chip|14520|o4741g0237o| o4741g0240o.chip.14522.36.1|2009-04-15T03:02:14Z|notset |1|chip|14522|o4741g0240o| o4741g0239o.chip.14524.38.1|2009-04-15T03:02:14Z|notset |1|chip|14524|o4741g0239o| o4741g0241o.chip.14523.37.1|2009-04-15T03:02:15Z|notset |1|chip|14523|o4741g0241o| o4741g0238o.chip.14521.35.1|2009-04-15T04:03:19Z|notset |1|chip|14521|o4741g0238o|
Note: the fileset type for IPP distribution bundles has not yet been defined. (notset in this example)
Fileset Names
Runs that are the result of a particular exposure
$exp_name.$stage.$stage_id.$dist_id.$prod_id
$exp_name is rawExp.exp_name $stage is raw, chip, camera, fake, warp, or diff $dist_id is distRun.dist_id (This table is internal to the server side of the distribution system) $prod_id is rcDSProduct.prod_id (This table is internal to the server side of the distribution system)
Examples
o4741g0239o.raw.36625.27.1 o4741g0239o.chip.11955.2.1
For stackRun distributions the fileset name is
$stack.$stack_id.$dist_id.$prod_id
for example
stack.9554.42.2
The name of detrend filesets is TBD.
Contents of the fileset
1) An mdc file containing the database information for the run with the name
dbinfo.$stage.$stage_id.mdc
for example
dbinfo.chip.11955.mdc
This file is the output from the -exportrun mode of the stage's ippTool (regtool, chiptool, etc). If the run is exported with the -clean option when the run is in full state the state and data state are changed from 'full' to 'cleaned'
XXX: Should we be doing something with the nebulous paths in this file here? Here I assume this is the client's job.
2) A set of gzipped tar files containing the files for a given component of the run.
For stages raw, chip, camera, and fake the name of the file for each component is
basename($component_path_base).$class_id.tgz
and for stages warp, diff, and stack the filenames are
basename($component_path_base).tgz
For example the files for chip XY01 for chipRun 11955 the path_base is
neb://anyhost/md01_20081002/o4741g0239o.36625/o4741g0239o.36625.ch.11955
so the name of the tarfile is
o4741g0239o.36625.ch.11955.XY05.tgz
The filenames in the tarfile are the basenames of the nebulous paths. For example
% tar tf o4741g0239o.36625.ch.11955.XY05.tgz ./ ./o4741g0239o.36625.ch.11955.XY05.ppImage.mdc ./o4741g0239o.36625.ch.11955.XY05.b2.fits ./o4741g0239o.36625.ch.11955.XY05.stats ./o4741g0239o.36625.ch.11955.XY05.cmf ./o4741g0239o.36625.ch.11955.XY05.trace ./o4741g0239o.36625.ch.11955.XY05.log ./o4741g0239o.36625.ch.11955.XY05.mdl.fits ./o4741g0239o.36625.ch.11955.XY05.b1.fits
Notes on the process of importing the data into the mirror system The client system runs a task that queries the ipp distribution data store product for new filesets
dsproductls --uri $uri --last_fileset $last_fileset_processed
For each fileset to be downloaded
dsget each file in the fileset to local temporary storage
parse the dbinfo file to find the workdirs and path_base foreach component
edit paths as necessary for the mirror installation foreach file insert into nebulous
import the dbinfo into the database using xxxtool -importrun
ISSUES/Question?
