Persist species match index

17518
Reporter: mdoering
Assignee: mdoering
Type: NewFeature
Summary: Persist species match index
Priority: Major
Resolution: Fixed
Status: Closed
Created: 2015-03-24 15:51:10.544
Updated: 2015-03-26 18:44:43.116
Resolved: 2015-03-26 18:44:43.087
        
Description: The species match webservice (aka "nub lookup) builds a new lucene index in memory on startup. This takes 10-20 minutes depending on the environment while the backbone does not change that often.

Separate out an nub index builder and read a lucene index from the filesystem when the nub-ws starts up saves the entire startup time. Test if performance is still suitable for heavy indexing]]>
    

Author: mdoering@gbif.org
Comment: https://github.com/gbif/checklistbank/commit/19465f79ce8554786a788153e65aec195f956af1
Created: 2015-03-26 17:50:46.345
Updated: 2015-03-26 17:50:46.345

Author: mdoering@gbif.org
Created: 2015-03-26 18:00:03.606
Updated: 2015-03-26 18:00:03.606
        
dev and uat configs have been changed to use checklistbank.nub.indexDir=/usr/local/gbif/services/checklistbank-nub-ws/nubidx
And the execution properties a reduced heap of 1g only