[
https://issues.apache.org/jira/browse/GEODE-3928?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dan Smith updated GEODE-3928:
-----------------------------
Description:
After GEODE-3030 is complete, we will have an internal API to create an index
on an existing region.
After someone calls that API on all members that have the user's region we
should index all of the data in the existing region.
Acceptance:
After calling the API to add the index to an existing region, at some point in
the future calling a query on the region should include all of the entries that
were in the existing region.
Implementation Details:
We think this should be implemented by modifying computeRepository so that it
goes through the following steps:
# If there is no COMPLETE file in the fileAndChunkRegion for this bucket
## Iterate over the contents of the user region and add them to the lucene index
## Write a COMPLETE file to the fileAndChunkRegion
## Return the IndexRepository
# If there is a COMPLETE file, just return the IndexRepository.
Note: When this task is complete, it's possible queries may block until the
indexing is complete. We will address that issue in GEODE-3926
was:When GEODE-3925 is implemented, if COMPLETE file is not present in the
fileAndChunkRegion, the computeRespository method should iterate the existing
data (bucket level) in the data region and add it into the lucene index.
> After calling Internal API to create lucene index after region is created,
> data in the region should be included in the lucene index
> ------------------------------------------------------------------------------------------------------------------------------------
>
> Key: GEODE-3928
> URL: https://issues.apache.org/jira/browse/GEODE-3928
> Project: Geode
> Issue Type: Sub-task
> Components: lucene
> Reporter: Jason Huynh
> Priority: Major
>
> After GEODE-3030 is complete, we will have an internal API to create an index
> on an existing region.
> After someone calls that API on all members that have the user's region we
> should index all of the data in the existing region.
> Acceptance:
> After calling the API to add the index to an existing region, at some point
> in the future calling a query on the region should include all of the entries
> that were in the existing region.
> Implementation Details:
> We think this should be implemented by modifying computeRepository so that it
> goes through the following steps:
> # If there is no COMPLETE file in the fileAndChunkRegion for this bucket
> ## Iterate over the contents of the user region and add them to the lucene
> index
> ## Write a COMPLETE file to the fileAndChunkRegion
> ## Return the IndexRepository
> # If there is a COMPLETE file, just return the IndexRepository.
> Note: When this task is complete, it's possible queries may block until the
> indexing is complete. We will address that issue in GEODE-3926
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
