[
https://issues.apache.org/jira/browse/HIVE-8684?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14191743#comment-14191743
]
Philipp Schuegerl commented on HIVE-8684:
-----------------------------------------
Potentially related to the already fixed 2758 (could have missed that ALTER
TABLE ... SET statement)
> Metastore caching too aggresive for partition location change
> -------------------------------------------------------------
>
> Key: HIVE-8684
> URL: https://issues.apache.org/jira/browse/HIVE-8684
> Project: Hive
> Issue Type: Bug
> Components: Metastore
> Affects Versions: 0.12.0
> Reporter: Philipp Schuegerl
>
> In a setup in which multiple metastores talk to a single MySQL DB, a
> partition location gets cached by the metastore although caching is turned
> off ("datanucleus.cache.level2" is set to "false").
> Reproduction steps:
> * Create a table <table> and partition <partition> with location X
> * Connect to hive with metastore A
> * Call ALTER TABLE <table> PARTITION (<partition>) SET LOCATION Y ... DESC
> FORMATTED <table> PARTITION (<partition>) will show location Y
> * Connect to hive with metastore B
> * DESC FORMATTED <table> PARTITION (<partition>) will still show location X
> A workaround for this problem is to DROP the partition and recreate it (which
> correctly propagates to all metastores). Only the SET statement is affected
> by the bug.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
