Hi Kyle:
I will try adding the object hash to the name to guarantee a unique id
for quartz.
The main concern is to get readable logs in
content/thredds/logs/featureCollectionScan.log
From that POV, I would recommend that unique names are used. One can
also put a name on the collection element, which will be used if its
present. However, Im not doing that on motherlode, because there's too
many damn names.
Thanks for reporting this problem.
John
On 6/13/2012 9:17 AM, Kyle Wilcox wrote:
It appears Quartz is using the 'name' attribute on a featureCollection for a
unique ID internally. Since the 'name' attribute on a featureColllection does
not have to be unique within a THREDDS catalog, it is possible to break the job
scheduler.
Should Quartz be using the "path" of a dataset to uniquely identify an update job rather
than the "name"?
[2012-06-13T16:51:37.791+0200] ERROR thredds.inventory.CollectionUpdater:
cronExecutor failed to schedule startup Job for Config{name='Domain1',...}
org.quartz.ObjectAlreadyExistsException: Unable to store Job with name:
'Domain1' and group: 'UpdateCollection', because one already exists with this
identification.
Example config that will generate the error (two different datasets with the same
"name='Domain1'"). Live here: http://ocean.uca.es:8080/thredds/catalog.html
<dataset name="UCA2D">
<featureCollection name="Domain1" path="fmrc/uca2d/Domain1" featureType="FMRC"
harvest="true">
...
<collection spec="C:/ProductionData/UCA2D/Domain1/**/.*\.nc$"
recheckAfter="15 min"
olderThan="5 min"/>
<update startup="true" rescan="0 */12 * * * ? *" />
</featureCollection>
</dataset>
<dataset name="UCA3D">
<featureCollection name="Domain1" path="fmrc/uca3d/Domain1" featureType="FMRC"
harvest="true">
...
<collection spec="C:/ProductionData/UCA3D/Domain1/**/.*\.nc$"
recheckAfter="15 min"
olderThan="5 min"/>
<update startup="true" rescan="0 */12 * * * ? *" />
</featureCollection>
</dataset>
_______________________________________________
thredds mailing list
thredds@xxxxxxxxxxxxxxxx
For list information or to unsubscribe, visit:
http://www.unidata.ucar.edu/mailing_lists/
