[
https://issues.apache.org/jira/browse/ASTERIXDB-33?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14961389#comment-14961389
]
Murtadha Hubail commented on ASTERIXDB-33:
------------------------------------------
Currently, an Asterix instance cannot be destroyed before it is stopped.
Closing the buffer cache is part of stopping an NC. I believe if an exception
happens during that operation, it should be reported.
The real problem is the failure during creating a secondary index which is not
reproducible in the current master. The problem could've been caused by a
missing close() or handleException() call, which unpin the pages of a search
cursor or bulk loader while bulk loading a secondary index. I have identified a
place in the current master where still can happen on
AbstractTreeIndexBulkLoader.end(). However, this will be fixed in the LSM
append only change:
https://asterix-gerrit.ics.uci.edu/#/c/255/21
> destroying an application fails if a previous job using b-trees failed
> ----------------------------------------------------------------------
>
> Key: ASTERIXDB-33
> URL: https://issues.apache.org/jira/browse/ASTERIXDB-33
> Project: Apache AsterixDB
> Issue Type: Bug
> Components: Hyracks
> Reporter: asterixdb-importer
> Assignee: Murtadha Hubail
> Priority: Minor
>
> destroying an application fails if a previous job using b-trees failed
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
