[
https://issues.apache.org/jira/browse/PHOENIX-1927?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14515140#comment-14515140
]
Samarth Jain commented on PHOENIX-1927:
---------------------------------------
The call() method for each task is executed and the underlying scanners created
almost immediately especially if there is no other traffic on the phoenix
client. Because we exhaust an iterator completely before moving on to the next
iterator, it is likely that the scanners will face these kind of lease
expirations especially when the data volume is large. So either we should delay
scanner creation (which is the easiest option) or have something similar to
RoundRobinIterator for UPSERT SELECT on client side. In fact, even having a
server side round robin executor won't solve it completely because the client
scanners themselves won't be called in a round robin fashion.
> hbase.regionserver.lease.period needs to be increased for immutable index
> population
> ------------------------------------------------------------------------------------
>
> Key: PHOENIX-1927
> URL: https://issues.apache.org/jira/browse/PHOENIX-1927
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 4.3.1
> Reporter: Mujtaba Chohan
>
> Even after increasing phoenix.query.timeoutMs,
> hbase.regionserver.lease.period needs to be set to a high value otherwise
> following exception stops immutable index population when data is present in
> base table:
> org.apache.phoenix.exception.PhoenixIOException: 88979ms passed since the
> last invocation, timeout is currently set to 60000
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
